MongoDB Manual

MongoDB Documentation
Release 2.4.3
MongoDB Documentation Project
May 09, 2013
Contents
I
1
Install MongoDB
Installation Guides 1.1 Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux 1.2 Install MongoDB on Ubuntu . . . . . . . . . . . . . . . . . . . . . 1.3 Install MongoDB on Debian . . . . . . . . . . . . . . . . . . . . . 1.4 Install MongoDB on Linux . . . . . . . . . . . . . . . . . . . . . 1.5 Install MongoDB on OS X . . . . . . . . . . . . . . . . . . . . . . 1.6 Install MongoDB on Windows . . . . . . . . . . . . . . . . . . . . 1.7 Install MongoDB Enterprise . . . . . . . . . . . . . . . . . . . . . 1.8 Getting Started with MongoDB . . . . . . . . . . . . . . . . . . . Release Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
3 3 6 9 11 13 16 20 21 29
II
3
Administration
Run-time Database Conguration 3.1 Congure the Database . . . . . . . . . . . . . . . . . 3.2 Security Considerations . . . . . . . . . . . . . . . . 3.3 Replication and Sharding Conguration . . . . . . . . 3.4 Run Multiple Database Instances on the Same System 3.5 Diagnostic Congurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
35 35 36 37 38 38 41 41 43 54 61 61 62 63 64 71 71 73 75
Backup and Recovery Operations for MongoDB 4.1 Backup Strategies for MongoDB Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Backup and Recovery Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Backup and Restore Sharded Clusters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data Center Awareness 5.1 Operational Segregation in MongoDB Operations and Deployments 5.2 Tag Aware Sharding . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Administer and Manage Shard Tags . . . . . . . . . . . . . . . . . 5.4 Deploy a Geographically Distributed Replica Set . . . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
Journaling 6.1 Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Journaling Internals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Connect to MongoDB with SSL
7.1 7.2 8
Congure mongod and mongos for SSL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SSL Conguration for Clients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
75 77 81 81 82 83 85 85 86 87 89 89 89 91 91 94 94 97 97
Monitor MongoDB with SNMP 8.1 Prerequisites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Congure SNMP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Manage mongod Processes 9.1 Start mongod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Stop mongod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Sending a UNIX INT or TERM Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10 Rotate Log Files 10.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Monitoring for MongoDB 11.1 Monitoring Tools . . . . . . . . 11.2 Process Logging . . . . . . . . 11.3 Diagnosing Performance Issues 11.4 Replication and Monitoring . . 11.5 Sharding and Monitoring . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
12 Analyze Performance of Database Operations 12.1 Proling Levels . . . . . . . . . . . . . . . . . . . . . 12.2 Enable Database Proling and Set the Proling Level . 12.3 View Proler Data . . . . . . . . . . . . . . . . . . . 12.4 Proler Overhead . . . . . . . . . . . . . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
99 . 99 . 99 . 101 . 102
13 Import and Export MongoDB Data 105 13.1 Data Type Fidelity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 13.2 Data Import and Export and Backups Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 13.3 Human Intelligible Import/Export Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 14 Linux ulimit Settings 109 14.1 Resource Utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 14.2 Review and Set Resource Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 14.3 Recommended Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 15 Production Notes 15.1 Backups . . . . . . . . . . . . . . . . . 15.2 Networking . . . . . . . . . . . . . . . . 15.3 MongoDB on Linux . . . . . . . . . . . 15.4 Readahead . . . . . . . . . . . . . . . . 15.5 MongoDB on Virtual Environments . . . 15.6 Disk and Storage Systems . . . . . . . . 15.7 Hardware Requirements and Limitations 15.8 Performance Monitoring . . . . . . . . . 15.9 Production Checklist . . . . . . . . . . . 113 113 113 113 114 114 114 115 116 116
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
16 Use Database Commands 121 16.1 Database Command Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 16.2 Issue Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 16.3 admin Database Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
ii
16.4 Command Responses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 17 MongoDB Tutorials 17.1 Getting Started . . . . . . . 17.2 Administration . . . . . . . 17.3 Development Patterns . . . 17.4 Application Development . 17.5 Text Search Patterns . . . . 17.6 Data Modeling Patterns . . 17.7 MongoDB Use Case Studies 123 123 123 125 125 126 126 126
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
III
Security
127
18 Security Concepts and Strategies 131 18.1 Security Practices and Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 18.2 Access Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 19 Tutorials 137 19.1 Network Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 19.2 Access Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 20 Reference 20.1 User Privilege Roles in MongoDB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.2 system.users Privilege Documents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20.3 Password Hashing Insecurity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 155 159 162
IV
Core MongoDB Operations (CRUD)
163
21 Read and Write Operations in MongoDB 167 21.1 Read Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 21.2 Write Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 21.3 Write Concern Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 22 Fundamental Concepts for Document Databases 22.1 BSON Documents . . . . . . . . . . . . . . 22.2 ObjectId . . . . . . . . . . . . . . . . . . . 22.3 GridFS . . . . . . . . . . . . . . . . . . . . 22.4 Database References . . . . . . . . . . . . . 23 CRUD Operations for MongoDB 23.1 Create . . . . . . . . . . . . 23.2 Read . . . . . . . . . . . . 23.3 Update . . . . . . . . . . . 23.4 Delete . . . . . . . . . . . . 187 187 194 196 197 201 201 209 219 227
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
Data Modeling
229
24 Background 233 24.1 Data Modeling Considerations for MongoDB Applications . . . . . . . . . . . . . . . . . . . . . . . 233 25 Data Modeling Patterns 239 25.1 Model Embedded One-to-One Relationships Between Documents . . . . . . . . . . . . . . . . . . . 239 25.2 Model Embedded One-to-Many Relationships Between Documents . . . . . . . . . . . . . . . . . . 240 iii
25.3 25.4 25.5 25.6 25.7 25.8 25.9 25.10
Model Referenced One-to-Many Relationships Between Documents Model Data for Atomic Operations . . . . . . . . . . . . . . . . . Model Tree Structures with Parent References . . . . . . . . . . . Model Tree Structures with Child References . . . . . . . . . . . . Model Tree Structures with an Array of Ancestors . . . . . . . . . Model Tree Structures with Materialized Paths . . . . . . . . . . . Model Tree Structures with Nested Sets . . . . . . . . . . . . . . . Model Data to Support Keyword Search . . . . . . . . . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
241 243 244 244 245 246 247 247
VI
Aggregation
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
251
255 255 255 256 257 259 260
26 Aggregation Framework 26.1 Overview . . . . . . . . 26.2 Framework Components 26.3 Use . . . . . . . . . . . 26.4 Optimizing Performance 26.5 Sharded Operation . . . 26.6 Limitations . . . . . . .
27 Aggregation Framework Examples 261 27.1 Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 27.2 Aggregations using the Zip Code Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 27.3 Aggregation with User Preference Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 28 Aggregation Framework Reference 28.1 $add (aggregation) . . . . . . . 28.2 $addToSet (aggregation) . . . . 28.3 $and (aggregation) . . . . . . . 28.4 $avg (aggregation) . . . . . . . 28.5 $cmp (aggregation) . . . . . . . 28.6 $concat (aggregation) . . . . . 28.7 $cond (aggregation) . . . . . . 28.8 $dayOfMonth (aggregation) . . 28.9 $dayOfWeek (aggregation) . . . 28.10 $dayOfYear (aggregation) . . . 28.11 $divide (aggregation) . . . . . . 28.12 $eq (aggregation) . . . . . . . . 28.13 $rst (aggregation) . . . . . . . 28.14 $geoNear (aggregation) . . . . 28.15 $group (aggregation) . . . . . . 28.16 $gt (aggregation) . . . . . . . . 28.17 $gte (aggregation) . . . . . . . 28.18 $hour (aggregation) . . . . . . 28.19 $ifNull (aggregation) . . . . . . 28.20 $last (aggregation) . . . . . . . 28.21 $limit (aggregation) . . . . . . 28.22 $lt (aggregation) . . . . . . . . 28.23 $lte (aggregation) . . . . . . . . 28.24 $match (aggregation) . . . . . . 28.25 $max (aggregation) . . . . . . . 28.26 $millisecond (aggregation) . . . 28.27 $min (aggregation) . . . . . . . 28.28 $minute (aggregation) . . . . . 271 272 272 272 272 272 272 275 275 275 276 276 276 276 276 278 279 280 280 280 280 280 281 281 281 282 282 283 284
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
iv
28.29 28.30 28.31 28.32 28.33 28.34 28.35 28.36 28.37 28.38 28.39 28.40 28.41 28.42 28.43 28.44 28.45 28.46 28.47 28.48 28.49 28.50
$mod (aggregation) . . . . $month (aggregation) . . $multiply (aggregation) . $ne (aggregation) . . . . . $not (aggregation) . . . . $or (aggregation) . . . . . $project (aggregation) . . $push (aggregation) . . . $second (aggregation) . . $skip (aggregation) . . . . $sort (aggregation) . . . . $strcasecmp (aggregation) $substr (aggregation) . . . $subtract (aggregation) . . $sum (aggregation) . . . . $toLower (aggregation) . $toUpper (aggregation) . . $unwind (aggregation) . . $week (aggregation) . . . $year (aggregation) . . . . Pipeline . . . . . . . . . . Expressions . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
284 284 284 285 285 285 285 287 287 287 287 289 289 289 289 289 290 290 291 291 291 300
29 SQL to Aggregation Framework Mapping Chart 307 29.1 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307 30 Map-Reduce 30.1 Examples . . . . . . . . . . . . . . . . . 30.2 Temporary Collection . . . . . . . . . . 30.3 Concurrency . . . . . . . . . . . . . . . 30.4 Sharded Cluster . . . . . . . . . . . . . 30.5 Troubleshooting Map-Reduce Operations 311 311 316 316 316 317
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
31 Simple Aggregation Methods and Commands 323 31.1 Count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 31.2 Distinct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 31.3 Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
VII
Indexes
325
32 Index Concepts 329 32.1 Indexing Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329 33 Indexing Strategies for Applications 341 33.1 Indexing Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341 34 Index Tutorials 347 34.1 Indexing Operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347 35 Geospatial Indexing 357 35.1 Geospatial Indexes and Queries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357 36 Text Indexing 371 36.1 Text Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
VIII
Replication
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
383
387 387 395 398 408 412
37 Replica Set Use and Operation 37.1 Replica Set Fundamental Concepts . . . . . . . . . . . . . . . . . . . . . . 37.2 Replica Set Architectures and Deployment Patterns . . . . . . . . . . . . . . 37.3 Replica Set Considerations and Behaviors for Applications and Development 37.4 Replica Set Internals and Behaviors . . . . . . . . . . . . . . . . . . . . . . 37.5 Master Slave Replication . . . . . . . . . . . . . . . . . . . . . . . . . . . .
38 Replica Set Tutorials and Procedures 419 38.1 Replica Set Administration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419 39 Replica Set Reference Material 463 39.1 Replica Set Conguration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463 39.2 Replica Set Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468 39.3 Replica Set Features and Version Compatibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
IX
Sharding
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
481
485 485 487 490 492 493
40 Sharding Concepts 40.1 Sharded Cluster Overview . . . . . . . 40.2 Sharded Cluster Architectures . . . . . 40.3 Query Routing in Sharded Clusters . . 40.4 Security Practices for Sharded Clusters 40.5 Sharded Cluster Internals . . . . . . .
41 Administration 503 41.1 Sharded Cluster Administration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503 42 Reference 537 42.1 Sharding Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537 42.2 Cong Database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
Application Development
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
551
555 555 555 558 560 561 563 563 568 571 572 575 577
43 Development Considerations 43.1 MongoDB Drivers and Client Libraries . 43.2 Optimization Strategies for MongoDB . 43.3 Capped Collections . . . . . . . . . . . . 43.4 Server-side JavaScript . . . . . . . . . . 43.5 Store a JavaScript Function on the Server
44 Application Design Patterns for MongoDB 44.1 Perform Two Phase Commits . . . . . . . . . . . . . 44.2 Create Tailable Cursor . . . . . . . . . . . . . . . . . 44.3 Isolate Sequence of Operations . . . . . . . . . . . . 44.4 Create an Auto-Incrementing Sequence Field . . . . . 44.5 Limit Number of Elements in an Array after an Update 44.6 Expire Data from Collections by Setting TTL . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
XI
The mongo Shell
579
583
45 Getting Started with the mongo Shell vi
45.1 45.2 45.3 45.4 45.5 45.6
Start the mongo Shell . . . . . . . . . . . Executing Queries . . . . . . . . . . . . . Print . . . . . . . . . . . . . . . . . . . . Use a Custom Prompt . . . . . . . . . . . Use an External Editor in the mongo Shell Exit the Shell . . . . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
583 584 584 585 586 586 587 587 588 588 589 591 591 591 591 592 592 593
46 Data Types in the mongo Shell 46.1 Date . . . . . . . . . . . . 46.2 ObjectId . . . . . . . . . 46.3 NumberLong . . . . . . . 46.4 NumberInt . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
47 Access the mongo Shell Help Information 47.1 Command Line Help . . . . . . . . . 47.2 Shell Help . . . . . . . . . . . . . . 47.3 Database Help . . . . . . . . . . . . 47.4 Collection Help . . . . . . . . . . . . 47.5 Cursor Help . . . . . . . . . . . . . 47.6 Type Help . . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
48 Write Scripts for the mongo Shell 595 48.1 Opening New Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595 48.2 Scripting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596 49 mongo Shell Quick Reference 49.1 mongo Shell Command History . 49.2 Command Line Options . . . . . 49.3 Command Helpers . . . . . . . . 49.4 Basic Shell JavaScript Operations 49.5 Keyboard Shortcuts . . . . . . . 49.6 Queries . . . . . . . . . . . . . . 49.7 Error Checking Methods . . . . . 49.8 Administrative Command Helpers 49.9 Opening Additional Connections 49.10 Miscellaneous . . . . . . . . . . 49.11 Additional Resources . . . . . . . 597 597 597 597 598 599 600 603 603 603 604 604
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
XII
Use Cases
605
50 Operational Intelligence 609 50.1 Storing Log Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609 50.2 Pre-Aggregated Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619 50.3 Hierarchical Aggregation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628 51 Product Data Management 51.1 Product Catalog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.2 Inventory Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.3 Category Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637 637 645 651
52 Content Management Systems 659 52.1 Metadata and Asset Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659 52.2 Storing Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 666
vii
53 Python Application Development 677 53.1 Write a Tumblelog Application with Django MongoDB Engine . . . . . . . . . . . . . . . . . . . . 677 53.2 Write a Tumblelog Application with Flask and MongoEngine . . . . . . . . . . . . . . . . . . . . . 689
XIII
Frequently Asked Questions

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
707
709 709 710 710 710 710 710 711 711 711 711 712 712 712 712 713 713 714 714 714 714 715 715 715 715 716 716 717 718 719 720 720 721 721 722 723 723 723 724 724 724
54 FAQ: MongoDB Fundamentals 54.1 What kind of database is MongoDB? . . . . . . . . . . . . . . . . . . . . . . 54.2 Do MongoDB databases have tables? . . . . . . . . . . . . . . . . . . . . . . 54.3 Do MongoDB databases have schemas? . . . . . . . . . . . . . . . . . . . . . 54.4 What languages can I use to work with MongoDB? . . . . . . . . . . . . . . . 54.5 Does MongoDB support SQL? . . . . . . . . . . . . . . . . . . . . . . . . . 54.6 What are typical uses for MongoDB? . . . . . . . . . . . . . . . . . . . . . . 54.7 Does MongoDB support transactions? . . . . . . . . . . . . . . . . . . . . . . 54.8 Does MongoDB require a lot of RAM? . . . . . . . . . . . . . . . . . . . . . 54.9 How do I congure the cache size? . . . . . . . . . . . . . . . . . . . . . . . 54.10 Does MongoDB require a separate caching layer for application-level caching? 54.11 Does MongoDB handle caching? . . . . . . . . . . . . . . . . . . . . . . . . 54.12 Are writes written to disk immediately, or lazily? . . . . . . . . . . . . . . . . 54.13 What language is MongoDB written in? . . . . . . . . . . . . . . . . . . . . . 54.14 What are the limitations of 32-bit versions of MongoDB? . . . . . . . . . . . 55 FAQ: MongoDB for Application Developers 55.1 What is a namespace in MongoDB? . . . . . . . . . . . . . . . . . 55.2 How do you copy all objects from one collection to another? . . . . 55.3 If you remove a document, does MongoDB remove it from disk? . 55.4 When does MongoDB write updates to disk? . . . . . . . . . . . . 55.5 How do I do transactions and locking in MongoDB? . . . . . . . . 55.6 How do you aggregate data with MongoDB? . . . . . . . . . . . . 55.7 Why does MongoDB log so many Connection Accepted events? 55.8 Does MongoDB run on Amazon EBS? . . . . . . . . . . . . . . . 55.9 Why are MongoDBs data les so large? . . . . . . . . . . . . . . 55.10 How do I optimize storage use for small documents? . . . . . . . . 55.11 When should I use GridFS? . . . . . . . . . . . . . . . . . . . . . 55.12 How does MongoDB address SQL or Query injection? . . . . . . . 55.13 How does MongoDB provide concurrency? . . . . . . . . . . . . . 55.14 What is the compare order for BSON types? . . . . . . . . . . . . 55.15 How do I query for elds that have null values? . . . . . . . . . . . 55.16 Are there any restrictions on the names of Collections? . . . . . . . 55.17 How do I isolate cursors from intervening write operations? . . . . 55.18 When should I embed documents within other documents? . . . . . 55.19 Can I manually pad documents to prevent moves during updates? .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
56 FAQ: The mongo Shell 56.1 How can I enter multi-line operations in the mongo shell? . . . . . . . . . . 56.2 How can I access different databases temporarily? . . . . . . . . . . . . . . 56.3 Does the mongo shell support tab completion and other keyboard shortcuts? 56.4 How can I customize the mongo shell prompt? . . . . . . . . . . . . . . . . 56.5 Can I edit long shell operations with an external text editor? . . . . . . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
57 FAQ: Concurrency 727 57.1 What type of locking does MongoDB use? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 727 57.2 How granular are locks in MongoDB? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 728 57.3 How do I see the status of locks on my mongod instances? . . . . . . . . . . . . . . . . . . . . . . 728 viii
57.4 57.5 57.6 57.7 57.8 57.9 57.10 57.11
Does a read or write operation ever yield the lock? . . . . . . . . . . . . . . Which operations lock the database? . . . . . . . . . . . . . . . . . . . . . Which administrative commands lock the database? . . . . . . . . . . . . . Does a MongoDB operation ever lock more than one database? . . . . . . . How does sharding affect concurrency? . . . . . . . . . . . . . . . . . . . . How does concurrency affect a replica set primary? . . . . . . . . . . . . . . How does concurrency affect secondaries? . . . . . . . . . . . . . . . . . . What kind of concurrency does MongoDB provide for JavaScript operations?
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
. . . . . . . .
728 728 729 730 730 730 730 730 731 732 732 732 732 732 733 733 733 733 734 734 734 734 734 735 735 735 735 735 735 736 736 736 736 736 737 737 737 739 739 739 740 740 740 740 741 741 741 741 741 742 742 742
58 FAQ: Sharding with MongoDB 58.1 Is sharding appropriate for a new deployment? . . . . . . . . . . . . . . . . . . . . . . . . . 58.2 How does sharding work with replication? . . . . . . . . . . . . . . . . . . . . . . . . . . . 58.3 Can I change the shard key after sharding a collection? . . . . . . . . . . . . . . . . . . . . . 58.4 What happens to unsharded collections in sharded databases? . . . . . . . . . . . . . . . . . 58.5 How does MongoDB distribute data across shards? . . . . . . . . . . . . . . . . . . . . . . . 58.6 What happens if a client updates a document in a chunk during a migration? . . . . . . . . . 58.7 What happens to queries if a shard is inaccessible or slow? . . . . . . . . . . . . . . . . . . . 58.8 How does MongoDB distribute queries among shards? . . . . . . . . . . . . . . . . . . . . . 58.9 How does MongoDB sort queries in sharded environments? . . . . . . . . . . . . . . . . . . 58.10 How does MongoDB ensure unique _id eld values when using a shard key other than _id? 58.11 Ive enabled sharding and added a second shard, but all the data is still on one server. Why? . 58.12 Is it safe to remove old les in the moveChunk directory? . . . . . . . . . . . . . . . . . . . 58.13 How does mongos use connections? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58.14 Why does mongos hold connections open? . . . . . . . . . . . . . . . . . . . . . . . . . . . 58.15 Where does MongoDB report on connections used by mongos? . . . . . . . . . . . . . . . . 58.16 What does writebacklisten in the log mean? . . . . . . . . . . . . . . . . . . . . . . . 58.17 How should administrators deal with failed migrations? . . . . . . . . . . . . . . . . . . . . 58.18 What is the process for moving, renaming, or changing the number of cong servers? . . . . 58.19 When do the mongos servers detect cong server changes? . . . . . . . . . . . . . . . . . . 58.20 Is it possible to quickly update mongos servers after updating a replica set conguration? . . 58.21 What does the maxConns setting on mongos do? . . . . . . . . . . . . . . . . . . . . . . . 58.22 How do indexes impact queries in sharded systems? . . . . . . . . . . . . . . . . . . . . . . 58.23 Can shard keys be randomly generated? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58.24 Can shard keys have a non-uniform distribution of values? . . . . . . . . . . . . . . . . . . . 58.25 Can you shard on the _id eld? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58.26 Can shard key be in ascending order, like dates or timestamps? . . . . . . . . . . . . . . . . . 58.27 What do moveChunk commit failed errors mean? . . . . . . . . . . . . . . . . . . . 58.28 How does draining a shard affect the balancing of uneven chunk distribution? . . . . . . . . . 59 FAQ: Replica Sets and Replication in MongoDB 59.1 What kinds of replication does MongoDB support? . . . . . . . . . . . . . 59.2 What do the terms primary and master mean? . . . . . . . . . . . . . 59.3 What do the terms secondary and slave mean? . . . . . . . . . . . . . 59.4 How long does replica set failover take? . . . . . . . . . . . . . . . . . . . 59.5 Does replication work over the Internet and WAN connections? . . . . . . 59.6 Can MongoDB replicate over a noisy connection? . . . . . . . . . . . . 59.7 What is the preferred replication method: master/slave or replica sets? . . . 59.8 What is the preferred replication method: replica sets or replica pairs? . . . 59.9 Why use journaling if replication already provides data redundancy? . . . 59.10 Are write operations durable if write concern does not acknowledge writes? 59.11 How many arbiters do replica sets need? . . . . . . . . . . . . . . . . . . 59.12 What information do arbiters exchange with the rest of the replica set? . . 59.13 Which members of a replica set vote in elections? . . . . . . . . . . . . . 59.14 Do hidden members vote in replica set elections? . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
ix
59.15 Is it normal for replica set members to use different amounts of disk space? . . . . . . . . . . . . . . 743 60 FAQ: MongoDB Storage 60.1 What are memory mapped les? . . . . . . . . . . . . . . . . . . . . . . . 60.2 How do memory mapped les work? . . . . . . . . . . . . . . . . . . . . 60.3 How does MongoDB work with memory mapped les? . . . . . . . . . . 60.4 What are page faults? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60.5 What is the difference between soft and hard page faults? . . . . . . . . . 60.6 What tools can I use to investigate storage use in MongoDB? . . . . . . . 60.7 What is the working set? . . . . . . . . . . . . . . . . . . . . . . . . . . . 60.8 Why are the les in my data directory larger than the data in my database? 60.9 How can I check the size of a collection? . . . . . . . . . . . . . . . . . . 60.10 How can I check the size of indexes? . . . . . . . . . . . . . . . . . . . . 60.11 How do I know when the server runs out of disk space? . . . . . . . . . . 61 FAQ: Indexes 61.1 Should you run ensureIndex() after every insert? . . . . . . 61.2 How do you know what indexes exist in a collection? . . . . . . . 61.3 How do you determine the size of an index? . . . . . . . . . . . . 61.4 What happens if an index does not t into RAM? . . . . . . . . . 61.5 How do you know what index a query used? . . . . . . . . . . . 61.6 How do you determine what elds to index? . . . . . . . . . . . 61.7 How do write operations affect indexes? . . . . . . . . . . . . . . 61.8 Will building a large index affect database performance? . . . . . 61.9 Can I use index keys to constrain query matches? . . . . . . . . . 61.10 Using $ne and $nin in a query is slow. Why? . . . . . . . . . . 61.11 Can I use a multi-key index to support a query for a whole array? 61.12 How can I effectively use indexes strategy for attribute lookups? . 745 745 745 746 746 746 746 746 747 748 748 748 751 751 751 752 752 752 752 752 752 753 753 753 753 755 755 756 756 757
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
62 FAQ: MongoDB Diagnostics 62.1 Where can I nd information about a mongod process that stopped running unexpectedly? 62.2 Does TCP keepalive time affect sharded clusters and replica sets? . . . . . . . . . . . . 62.3 Memory Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62.4 Sharded Cluster Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
XIV
Reference
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
761
763 763 807 917 1016
63 MongoDB Interface 63.1 Query, Update and Projection Operators 63.2 Database Commands . . . . . . . . . . 63.3 mongo Shell Methods . . . . . . . . . 63.4 SQL to MongoDB Mapping Chart . . .
64 Architecture and Components 1021 64.1 MongoDB Package Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1021 65 Internal Metadata and Reporting 65.1 The local Database . . . . 65.2 System Collections . . . . . . 65.3 Database Proler Output . . . 65.4 Exit Codes and Statuses . . . 1097 1097 1099 1100 1103
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
66 General Reference 1105 66.1 MongoDB Limits and Thresholds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1105
66.2 66.3 66.4 66.5 66.6
Connection String URI Format MongoDB Extended JSON . . Database References . . . . . . GridFS Reference . . . . . . . Glossary . . . . . . . . . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
1108 1113 1115 1117 1119
XV
Release Notes
1131
67 Current Stable Release 1135 67.1 Release Notes for MongoDB 2.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1135 68 Previous Stable Releases 68.1 Release Notes for MongoDB 2.2 . . 68.2 Release Notes for MongoDB 2.0 . . 68.3 Release Notes for MongoDB 1.8 . . 68.4 Release Notes for MongoDB 1.6 . . 68.5 Release Notes for MongoDB 1.4 . . 68.6 Release Notes for MongoDB 1.2.x . 1157 1157 1167 1173 1178 1180 1182
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
69 Other MongoDB Release Notes 1185 69.1 Default Write Concern Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1185 70 Version Numbers 1187
XVI
About MongoDB Documentation
1189
1193 1195 1197 1199
71 License 72 Editions 73 Version and Revisions 74 Report an Issue or Make a Change Request
75 Contribute to the Documentation 1201 75.1 MongoDB Manual Translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1201 75.2 About the Documentation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1202
xi
xii
Part I
Install MongoDB
CHAPTER 1
Installation Guides
MongoDB runs on most platforms, and supports 32-bit and 64-bit architectures. 10gen, the MongoDB makers, provides both binaries and packages. Choose your platform below:
1.1 Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux

1.1.1 Synopsis
This tutorial outlines the basic installation process for deploying MongoDB on Red Hat Enterprise Linux, CentOS Linux, Fedora Linux and related systems. This procedure uses .rpm packages as the basis of the installation. 10gen publishes packages of the MongoDB releases as .rpm packages for easy installation and management for users of CentOS, Fedora and Red Hat Enterprise Linux systems. While some of these distributions include their own MongoDB packages, the 10gen packages are generally more up to date. This tutorial includes: an overview of the available packages, instructions for conguring the package manager, the process install packages from the 10gen repository, and preliminary MongoDB conguration and operation. See also: Additional installation tutorials: http://docs.mongodb.org/manual/tutorial/install-mongodb-on-debian-or-ubuntu-linux Install MongoDB on Debian (page 9) Install MongoDB on Ubuntu (page 6) Install MongoDB on Linux (page 11) Install MongoDB on OS X (page 13) Install MongoDB on Windows (page 16)
1.1.2 Package Options

The 10gen repository contains two packages: mongo-10gen-server This package contains the mongod (page 1021) and mongos (page 1032) daemons from the latest stable release and associated conguration and init scripts. Additionally, you can use this package to install daemons from a previous release (page 4) of MongoDB.
MongoDB Documentation, Release 2.4.3
mongo-10gen This package contains all MongoDB tools from the latest stable release. Additionally, you can use this package to install tools from a previous release (page 4) of MongoDB. Install this package on all production MongoDB hosts and optionally on other systems from which you may need to administer MongoDB systems.
1.1.3 Install MongoDB

Congure Package Management System (YUM) Create a http://docs.mongodb.org/manual/etc/yum.repos.d/10gen.repo le to hold information about your repository. If you are running a 64-bit system (recommended,) place the following conguration in http://docs.mongodb.org/manual/etc/yum.repos.d/10gen.repo le:
[10gen] name=10gen Repository baseurl=http://downloads-distro.mongodb.org/repo/redhat/os/x86_64 gpgcheck=0 enabled=1
If you are running a 32-bit system, which isnt recommended for production deployments, place the following conguration in http://docs.mongodb.org/manual/etc/yum.repos.d/10gen.repo le:
[10gen] name=10gen Repository baseurl=http://downloads-distro.mongodb.org/repo/redhat/os/i686 gpgcheck=0 enabled=1
Install Packages Issue the following command (as root or with sudo) to install the latest stable version of MongoDB and the associated tools:
yum install mongo-10gen mongo-10gen-server
When this command completes, you have successfully installed MongoDB! Manage Installed Versions You can use the mongo-10gen and mongo-10gen-server packages to install previous releases of MongoDB. To install a specic release, append the version number, as in the following example:
yum install mongo-10gen-2.2.3 mongo-10gen-server-2.2.3
This installs the mongo-10gen and mongo-10gen-server packages with the 2.2.3 release. You can specify any available version of MongoDB; however yum will upgrade the mongo-10gen and mongo-10gen-server packages when a newer version becomes available. Use the following pinning procedure to prevent unintended upgrades. To pin a package, add the following line to your http://docs.mongodb.org/manual/etc/yum.conf le:
exclude=mongo-10gen,mongo-10gen-server
Chapter 1. Installation Guides
1.1.4 Congure MongoDB

These packages congure MongoDB using the http://docs.mongodb.org/manual/etc/mongod.conf le in conjunction with the control script. You can nd the init script at http://docs.mongodb.org/manual/etc/rc.d/init.d/mongod. This MongoDB instance will store its data les in the http://docs.mongodb.org/manual/var/lib/mongo and its log les in http://docs.mongodb.org/manual/var/log/mongo, and run using the mongod user account. Note: If you change the user that runs the MongoDB process, you will need to modify the access control rights to the http://docs.mongodb.org/manual/var/lib/mongo and http://docs.mongodb.org/manual/var/log/mongo directories.
1.1.5 Control MongoDB

Warning: With the introduction of systemd in Fedora 15, the control scripts included in the packages available in the 10gen repository are not compatible with Fedora systems. A correction is forthcoming, see SERVER-7285 for more information, and in the mean time use your own control scripts or install using the procedure outlined in Install MongoDB on Linux (page 11).
Start MongoDB Start the mongod (page 1021) process by issuing the following command (as root, or with sudo):
service mongod start
You can verify that the mongod (page 1021) process has started successfully by checking the contents of the log le at http://docs.mongodb.org/manual/var/log/mongo/mongod.log. You may optionally, ensure that MongoDB will start following a system reboot, by issuing the following command (with root privileges:)
chkconfig mongod on
Stop MongoDB Stop the mongod (page 1021) process by issuing the following command (as root, or with sudo):
service mongod stop
Restart MongoDB You can restart the mongod (page 1021) process by issuing the following command (as root, or with sudo):
service mongod restart
Follow the state of this process by watching the output in the http://docs.mongodb.org/manual/var/log/mongo/mongod le to watch for errors or important messages from the server.
1.1. Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux
Control mongos As of the current release, there are no control scripts for mongos (page 1032). mongos (page 1032) is only used in sharding deployments and typically do not run on the same systems where mongod (page 1021) runs. You can use the mongodb script referenced above to derive your own mongos (page 1032) control script. SELinux Considerations You must SELinux to allow MongoDB to start on Fedora systems. Administrators have two options: enable access to the relevant ports (e.g. 27017) for SELinux. See Interfaces and Port Numbers (page 132) for more information on MongoDBs default ports. disable SELinux entirely. This requires a system reboot and may have larger implications for your deployment.
1.1.6 Using MongoDB

Among the tools included in the mongo-10gen package, is the mongo (page 1036) shell. You can connect to your MongoDB instance by issuing the following command at the system prompt:
mongo
This will connect to the database running on the localhost interface by default. At the mongo (page 1036) prompt, issue the following two commands to insert a record in the test collection of the (default) test database and then retrieve that document.
> db.test.save( { a: 1 } ) > db.test.find()
See also: mongo (page 1036) and mongo Shell Methods (page 917)
1.2 Install MongoDB on Ubuntu

1.2.1 Synopsis
This tutorial outlines the basic installation process for installing MongoDB on Ubuntu Linux systems. This tutorial uses .deb packages as the basis of the installation. 10gen publishes packages of the MongoDB releases as .deb packages for easy installation and management for users of Ubuntu systems. Although Ubuntu does include MongoDB packages, the 10gen packages are generally more up to date. This tutorial includes: an overview of the available packages, instructions for conguring the package manager, the process for installing packages from the 10gen repository, and preliminary MongoDB conguration and operation. Note: If you use an older Ubuntu that does not use Upstart, (i.e. any version before 9.10 Karmic) please follow the instructions on the Install MongoDB on Debian (page 9) tutorial. See also: Additional installation tutorials: Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux (page 3) Install MongoDB on Debian (page 9)
Install MongoDB on Linux (page 11) Install MongoDB on OS X (page 13) Install MongoDB on Windows (page 16)

The 10gen repository provides the mongodb-10gen package, which contains the latest stable release. Additionally you can install previous releases (page 7) of MongoDB. You cannot install this package concurrently with the mongodb, mongodb-server, or mongodb-clients packages provided by Ubuntu.

Congure Package Management System (APT) The Ubuntu package management tool (i.e. dpkg and apt) ensure package consistency and authenticity by requiring that distributors sign packages with GPG keys. Issue the following command to import the 10gen public GPG Key:
sudo apt-key adv --keyserver keyserver.ubuntu.com --recv 7F0CEB10
Create a http://docs.mongodb.org/manual/etc/apt/sources.list.d/10gen.list le and include the following line for the 10gen repository.
echo deb http://downloads-distro.mongodb.org/repo/ubuntu-upstart dist 10gen | sudo tee /etc/apt/sou
Now issue the following command to reload your repository:

sudo apt-get update
Install Packages Issue the following command to install the latest stable version of MongoDB:
sudo apt-get install mongodb-10gen
When this command completes, you have successfully installed MongoDB! Continue for conguration and start-up suggestions. Manage Installed Versions You can use the mongodb-10gen package to install previous versions of MongoDB. To install a specic release, append the version number to the package name, as in the following example:
apt-get install mongodb-10gen=2.2.3
This will install the 2.2.3 release of MongoDB. You can specify any available version of MongoDB; however apt-get will upgrade the mongodb-10gen package when a newer version becomes available. Use the following pinning procedure to prevent unintended upgrades. To pin a package, issue the following command at the system prompt to pin the version of MongoDB at the currently installed version:
1.2. Install MongoDB on Ubuntu
echo "mongodb-10gen hold" | dpkg --set-selections

These packages congure MongoDB using the http://docs.mongodb.org/manual/etc/mongodb.conf le in conjunction with the control script. You will nd the control script is at http://docs.mongodb.org/manual/etc/init.d/mongodb. This MongoDB instance will store its data les in the http://docs.mongodb.org/manual/var/lib/mongodb and its log les in http://docs.mongodb.org/manual/var/log/mongodb, and run using the mongodb user account. Note: If you change the user that runs the MongoDB process, you will need to modify the access control rights to the http://docs.mongodb.org/manual/var/lib/mongodb and http://docs.mongodb.org/manual/var/log/mongodb directories.
1.2.5 Controlling MongoDB

Starting MongoDB You can start the mongod (page 1021) process by issuing the following command:
sudo service mongodb start
You can verify that mongod (page 1021) has started successfully by checking the contents of the log le at http://docs.mongodb.org/manual/var/log/mongodb/mongodb.log. Stopping MongoDB As needed, you may stop the mongod (page 1021) process by issuing the following command:
sudo service mongodb stop
Restarting MongoDB You may restart the mongod (page 1021) process by issuing the following command:
sudo service mongodb restart
Controlling mongos As of the current release, there are no control scripts for mongos (page 1032). mongos (page 1032) is only used in sharding deployments and typically do not run on the same systems where mongod (page 1021) runs. You can use the mongodb script referenced above to derive your own mongos (page 1032) control script.
1.2.6 Using MongoDB

Among the tools included with the MongoDB package, is the mongo (page 1036) shell. You can connect to your MongoDB instance by issuing the following command at the system prompt:
mongo
This will connect to the database running on the localhost interface by default. At the mongo (page 1036) prompt, issue the following two commands to insert a record in the test collection of the (default) test database.
1.3 Install MongoDB on Debian

1.3.1 Synopsis
This tutorial outlines the basic installation process for installing MongoDB on Debian systems. This tutorial uses .deb packages as the basis of the installation. 10gen publishes packages of the MongoDB releases as .deb packages for easy installation and management for users of Debian systems. While some of these distributions include their own MongoDB packages, the 10gen packages are generally more up to date. This tutorial includes: an overview of the available packages, instructions for conguring the package manager, the process for installing packages from the 10gen repository, and preliminary MongoDB conguration and operation. Note: This tutorial applies to both Debian systems and versions of Ubuntu Linux prior to 9.10 Karmic which do not use Upstart. Other Ubuntu users will want to follow the Install MongoDB on Ubuntu (page 6) tutorial. See also: Additional installation tutorials: Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux (page 3) Install MongoDB on Ubuntu (page 6) Install MongoDB on Linux (page 11) Install MongoDB on OS X (page 13) Install MongoDB on Windows (page 16)

The 10gen repository provides the mongodb-10gen package, which contains the latest stable release. Additionally you can install previous releases (page 10) of MongoDB. You cannot install this package concurrently with the mongodb, mongodb-server, or mongodb-clients packages that your release of Debian may include.
1.3. Install MongoDB on Debian

Congure Package Management System (APT) The Debian package management tool (i.e. dpkg and apt) ensure package consistency and authenticity by requiring that distributors sign packages with GPG keys. Issue the following command to import the 10gen public GPG Key:
sudo apt-key adv --keyserver keyserver.ubuntu.com --recv 7F0CEB10
Create the http://docs.mongodb.org/manual/etc/apt/sources.list.d/10gen.list le and include a line for the 10gen repository with the following command:
echo deb http://downloads-distro.mongodb.org/repo/debian-sysvinit dist 10gen | sudo tee /etc/apt/so
Now issue the following command to reload your repository:

sudo apt-get update
Install Packages Issue the following command to install the latest stable version of MongoDB:
sudo apt-get install mongodb-10gen
When this command completes, you have successfully installed MongoDB! Manage Installed Versions You can use the mongodb-10gen package to install previous versions of MongoDB. To install a specic release, append the version number to the package name, as in the following example:
apt-get install mongodb-10gen=2.2.3
This will install the 2.2.3 release of MongoDB. You can specify any available version of MongoDB; however apt-get will upgrade the mongodb-10gen package when a newer version becomes available. Use the following pinning procedure to prevent unintended upgrades. To pin a package, issue the following command at the system prompt to pin the version of MongoDB at the currently installed version:
echo "mongodb-10gen hold" | dpkg --set-selections

These packages congure MongoDB using the http://docs.mongodb.org/manual/etc/mongodb.conf le in conjunction with the control script. You can nd the control script at http://docs.mongodb.org/manual/etc/init.d/mongodb. This MongoDB instance will store its data les in the http://docs.mongodb.org/manual/var/lib/mongodb and its log les in http://docs.mongodb.org/manual/var/log/mongodb, and run using the mongodb user account. Note: If you change the user that runs the MongoDB process, you will need to modify the access control rights to the http://docs.mongodb.org/manual/var/lib/mongodb and http://docs.mongodb.org/manual/var/log/mongodb directories.
10
1.3.5 Controlling MongoDB

Starting MongoDB Issue the following command to start mongod (page 1021):
sudo /etc/init.d/mongodb start
You can verify that mongod (page 1021) has started successfully by checking the contents of the log le at http://docs.mongodb.org/manual/var/log/mongodb/mongodb.log. Stopping MongoDB Issue the following command to stop mongod (page 1021):
sudo /etc/init.d/mongodb stop
Restarting MongoDB Issue the following command to restart mongod (page 1021):
sudo /etc/init.d/mongodb restart
Controlling mongos As of the current release, there are no control scripts for mongos (page 1032). mongos (page 1032) is only used in sharding deployments and typically do not run on the same systems where mongod (page 1021) runs. You can use the mongodb script referenced above to derive your own mongos (page 1032) control script.
1.3.6 Using MongoDB

Among the tools included with the MongoDB package, is the mongo (page 1036) shell. You can connect to your MongoDB instance by issuing the following command at the system prompt:
mongo
This will connect to the database running on the localhost interface by default. At the mongo (page 1036) prompt, issue the following two commands to insert a record in the test collection of the (default) test database.
1.4 Install MongoDB on Linux

1.4.1 Synopsis
10gen provides compiled versions of MongoDB for use on Linux that provides a simple option for users who cannot use packages. This tutorial outlines the basic installation of MongoDB using these compiled versions and an initial 1.4. Install MongoDB on Linux 11
usage guide. See also: Additional installation tutorials: Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux (page 3) Install MongoDB on Ubuntu (page 6) Install MongoDB on Debian (page 9) Install MongoDB on OS X (page 13) Install MongoDB on Windows (page 16)
1.4.2 Download MongoDB

Note: You should place the MongoDB binaries in a central location on the le system that is easy to access and control. Consider http://docs.mongodb.org/manual/opt or http://docs.mongodb.org/manual/usr/local/bin. In a terminal session, begin by downloading the latest release. In most cases you will want to download the 64-bit version of MongoDB.
curl http://downloads.mongodb.org/linux/mongodb-linux-x86_64-2.4.3.tgz > mongodb.tgz
If you need to run the 32-bit version, use the following command.
curl http://downloads.mongodb.org/linux/mongodb-linux-i686-2.4.3.tgz > mongodb.tgz
Once youve downloaded the release, issue the following command to extract the les from the archive:
tar -zxvf mongodb.tgz
Optional You may use the following command to copy the extracted folder into a more generic location.
cp -R -n mongodb-linux-????-??-??/ mongodb
You can nd the mongod (page 1021) binary, and the binaries all of the associated MongoDB utilities, in the bin/ directory within the extracted directory. Using MongoDB Before you start mongod (page 1021) for the rst time, you will need to create the data directory. By default, mongod (page 1021) writes data to the http://docs.mongodb.org/manual/data/db/ directory. To create this directory, use the following command:
mkdir -p /data/db
Note: Ensure that the system account that will run the mongod (page 1021) process has read and write permissions to this directory. If mongod (page 1021) runs under the mongo user account, issue the following command to change the owner of this folder:
12
chown mongo /data/db
If you use an alternate location for your data directory, ensure that this user can write to your chosen data path. You can specify, and create, an alternate path using the --dbpath (page 1023) option to mongod (page 1021) and the above command. The 10gen builds of MongoDB contain no control scripts or method to control the mongod (page 1021) process. You may wish to create control scripts, modify your path, and/or create symbolic links to the MongoDB programs in your http://docs.mongodb.org/manual/usr/local/bin or http://docs.mongodb.org/manual/usr/bin directory for easier use. For testing purposes, you can start a mongod (page 1021) directly in the terminal without creating a control script:
mongod --config /etc/mongod.conf
Note: The above command assumes that the mongod (page 1021) binary is accessible via your systems search path, and that you have created a default conguration le located at http://docs.mongodb.org/manual/etc/mongod.conf. Among the tools included with this MongoDB distribution, is the mongo (page 1036) shell. You can use this shell to connect to your MongoDB instance by issuing the following command at the system prompt:
./bin/mongo
Note: The ./bin/mongo command assumes that the mongo (page 1036) binary is in the bin/ sub-directory of the current directory. This is the directory into which you extracted the .tgz le. This will connect to the database running on the localhost interface by default. At the mongo (page 1036) prompt, issue the following two commands to insert a record in the test collection of the (default) test database and then retrieve that record:
1.5 Install MongoDB on OS X

Platform Support MongoDB only supports OS X versions 10.6 (Snow Leopard) and later. Changed in version 2.4.
1.5.1 Synopsis
This tutorial outlines the basic installation process for deploying MongoDB on Macintosh OS X systems. This tutorial provides two main methods of installing the MongoDB server (i.e. mongod (page 1021)) and associated tools: rst using the community package management tools, and second using builds of MongoDB provided by 10gen. See also: 1.5. Install MongoDB on OS X 13
Additional installation tutorials: Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux (page 3) Install MongoDB on Ubuntu (page 6) Install MongoDB on Debian (page 9) Install MongoDB on Linux (page 11) Install MongoDB on Windows (page 16)
1.5.2 Install with Package Management

Both community package management tools: Homebrew and MacPorts require some initial setup and conguration. This conguration is beyond the scope of this document. You only need to use one of these tools. If you want to use package management, and do not already have a system installed, Homebrew is typically easier and simpler to use. Homebrew Homebrew installs binary packages based on published formula. Issue the following command at the system shell to update the brew package manager:
brew update
Use the following command to install the MongoDB package into your Homebrew system.
brew install mongodb
Later, if you need to upgrade MongoDB, you can issue the following sequence of commands to update the MongoDB installation on your system:
brew update brew upgrade mongodb
MacPorts MacPorts distributes build scripts that allow you to easily build packages and their dependencies on your own system. The compilation process can take signicant period of time depending on your systems capabilities and existing dependencies. Issue the following command in the system shell:
port install mongodb
Using MongoDB from Homebrew and MacPorts The packages installed with Homebrew and MacPorts contain no control scripts or interaction with the systems process manager. If you have congured Homebrew and MacPorts correctly, including setting your PATH, the MongoDB applications and utilities will be accessible from the system shell. Start the mongod (page 1021) process in a terminal (for testing or development) or using a process management tool.
mongod
14
Then open the mongo (page 1036) shell by issuing the following command at the system prompt:
mongo
This will connect to the database running on the localhost interface by default. At the mongo (page 1036) prompt, issue the following two commands to insert a record in the test collection of the (default) test database and then retrieve that record.
1.5.3 Install from 10gen Builds

10gen provides compiled binaries of all MongoDB software compiled for OS X, which may provide a more straightforward installation process. Download MongoDB In a terminal session, begin by downloading the latest release. Use the following command at the system prompt:
curl http://downloads.mongodb.org/osx/mongodb-osx-x86_64-2.4.3.tgz > mongodb.tgz
Note: The mongod (page 1021) process will not run on older Macintosh computers with PowerPC (i.e. non-Intel) processors. Once youve downloaded the release, issue the following command to extract the les from the archive:
tar -zxvf mongodb.tgz
Optional You may use the following command to move the extracted folder into a more generic location.
mv -n mongodb-osx-[platform]-[version]/ /path/to/new/location/
Replace [platform] with i386 or x86_64 depending on your system and the version you downloaded, and [version] with 2.4 or the version of MongoDB that you are installing. You can nd the mongod (page 1021) binary, and the binaries all of the associated MongoDB utilities, in the bin/ directory within the archive. Using MongoDB from 10gen Builds Before you start mongod (page 1021) for the rst time, you will need to create the data directory. By default, mongod (page 1021) writes data to the http://docs.mongodb.org/manual/data/db/ directory. To create this directory, and set the appropriate permissions use the following commands:
sudo mkdir -p /data/db sudo chown id -u /data/db
1.5. Install MongoDB on OS X
15
You can specify an alternate path for data les using the --dbpath (page 1023) option to mongod (page 1021). The 10gen builds of MongoDB contain no control scripts or method to control the mongod (page 1021) process. You may wish to create control scripts, modify your path, and/or create symbolic links to the MongoDB programs in your http://docs.mongodb.org/manual/usr/local/bin directory for easier use. For testing purposes, you can start a mongod (page 1021) directly in the terminal without creating a control script:
mongod --config /etc/mongod.conf
Note: This command assumes that the mongod (page 1021) binary is accessible via your systems search path, and that you have created a default conguration le located at http://docs.mongodb.org/manual/etc/mongod.conf. Among the tools included with this MongoDB distribution, is the mongo (page 1036) shell. You can use this shell to connect to your MongoDB instance by issuing the following command at the system prompt from inside of the directory where you extracted mongo (page 1036):
./bin/mongo
Note: The ./bin/mongo command assumes that the mongo (page 1036) binary is in the bin/ sub-directory of the current directory. This is the directory into which you extracted the .tgz le. This will connect to the database running on the localhost interface by default. At the mongo (page 1036) prompt, issue the following two commands to insert a record in the test collection of the (default) test database and then retrieve that record:
1.6 Install MongoDB on Windows

1.6.1 Synopsis
This tutorial provides a method for installing and running the MongoDB server (i.e. mongod.exe (page 1041)) on the Microsoft Windows platform through the Command Prompt and outlines the process for setting up MongoDB as a Windows Service. Operating MongoDB with Windows is similar to MongoDB on other platforms. Most components share the same operational patterns.
1.6.2 Procedure
Important: If you are running any edition of Windows Server 2008 R2 or Windows 7, please install a hotx to resolve an issue with memory mapped les on Windows.
16
Download MongoDB for Windows Download the latest production release of MongoDB from the MongoDB downloads page. There are three builds of MongoDB for Windows: MongoDB for Windows Server 2008 R2 edition (i.e. 2008R2) only runs on Windows Server 2008 R2, Windows 7 64-bit, and newer versions of Windows. This build takes advantage of recent enhancements to the Windows Platform and cannot operate on older versions of Windows. MongoDB for Windows 64-bit runs on any 64-bit version of Windows newer than Windows XP, including Windows Server 2008 R2 and Windows 7 64-bit. MongoDB for Windows 32-bit runs on any 32-bit version of Windows newer than Windows XP. 32-bit versions of MongoDB are only intended for older systems and for use in testing and development systems. Changed in version 2.2: MongoDB does not support Windows XP. Please use a more recent version of Windows to use more recent releases of MongoDB. Note: Always download the correct version of MongoDB for your Windows system. The 64-bit versions of MongoDB will not work with 32-bit Windows. 32-bit versions of MongoDB are suitable only for testing and evaluation purposes and only support databases smaller than 2GB. You can nd the architecture of your version of Windows platform using the following command in the Command Prompt:
wmic os get osarchitecture
In Windows Explorer, nd the MongoDB download le, typically in the default Downloads directory. Extract the archive to C:\ by right clicking on the archive and selecting Extract All and browsing to C:\. Note: The folder name will be either:
C:\mongodb-win32-i386-[version]
Or:
C:\mongodb-win32-x86_64-[version]
In both examples, replace [version] with the version of MongoDB downloaded.
Set up the Environment Start the Command Prompt by selecting the Start Menu, then All Programs, then Accessories, then right click Command Prompt, and select Run as Administrator from the popup menu. In the Command Prompt, issue the following commands:
cd \ move C:\mongodb-win32-* C:\mongodb
Note: MongoDB is self-contained and does not have any other system dependencies. You can run MongoDB from any folder you choose. You may install MongoDB in any directory (e.g. D:\test\mongodb) MongoDB requires a data folder to store its les. The default location for the MongoDB data directory is C:\data\db. Create this folder using the Command Prompt. Issue the following command sequence:
1.6. Install MongoDB on Windows
17
md data md data\db
Note: You may specify an alternate path for \data\db with the dbpath (page 1081) setting for mongod.exe (page 1041), as in the following example:
C:\mongodb\bin\mongod.exe --dbpath d:\test\mongodb\data
If your path includes spaces, enclose the entire path in double quotations, for example:
C:\mongodb\bin\mongod.exe --dbpath "d:\test\mongo db data"
Start MongoDB To start MongoDB, execute from the Command Prompt:

C:\mongodb\bin\mongod.exe
This will start the main MongoDB database process. The waiting for connections message in the console output indicates that the mongod.exe process is running successfully. Note: Depending on the security level of your system, Windows will issue a Security Alert dialog box about blocking some features of C:\\mongodb\bin\mongod.exe from communicating on networks. All users should select Private Networks, such as my home or work network and click Allow access. For additional information on security and MongoDB, please read the Security Practices and Management (page 131) page. Warning: Do not allow mongod.exe (page 1041) to be accessible to public networks without running in Secure Mode (i.e. auth (page 1081).) MongoDB is designed to be run in trusted environments and the database does not enable authentication or Secure Mode by default. Connect to MongoDB using the mongo.exe (page 1036) shell. Open another Command Prompt and issue the following command:
C:\mongodb\bin\mongo.exe
Note: Executing the command start C:\mongodb\bin\mongo.exe will automatically start the mongo.exe shell in a separate Command Prompt window. The mongo.exe (page 1036) shell will connect to mongod.exe (page 1041) running on the localhost interface and port 27017 by default. At the mongo.exe (page 1036) prompt, issue the following two commands to insert a record in the test collection of the default test database and then retrieve that record:
See also: mongo (page 1036) and mongo Shell Methods (page 917). If you want to develop applications using .NET, see the documentation of C# and MongoDB for more information.
1.6.3 MongoDB as a Windows Service

New in version 2.0.
18
Setup MongoDB as a Windows Service, so that the database will start automatically following each reboot cycle. Note: mongod.exe (page 1041) added support for running as a Windows service in version 2.0, and mongos.exe (page 1042) added support for running as a Windows Service in version 2.1.1.
Congure the System You should specify two options when running MongoDB as a Windows Service: a path for the log output (i.e. logpath (page 1080)) and a conguration le (page 1078). 1. Create a specic directory for MongoDB log les:
md C:\mongodb\log
2. Create a conguration le for the logpath (page 1080) option for MongoDB in the Command Prompt by issuing this command:
echo logpath=C:\mongodb\log\mongo.log > C:\mongodb\mongod.cfg
While these optional steps are optional, creating a specic location for log les and using the conguration le are good practice. Note: Consider setting the logappend (page 1080) option. If you do not, mongod.exe (page 1041) will delete the contents of the existing log le when starting. Changed in version 2.2: The default logpath (page 1080) and logappend (page 1080) behavior changed in the 2.2 release.
Install and Run the MongoDB Service Run all of the following commands in Command Prompt with Administrative Privileges: 1. To install the MongoDB service:
C:\mongodb\bin\mongod.exe --config C:\mongodb\mongod.cfg --install
Modify the path to the mongod.cfg le as needed. For the --install (page 1041) option to succeed, you must specify a logpath (page 1080) setting or the --logpath (page 1022) run-time option. 2. To run the MongoDB service:
net start MongoDB
Note: If you wish to use an alternate path for your dbpath (page 1081) specify it in the cong le (e.g. C:\mongodb\mongod.cfg) on that you specied in the --install (page 1041) operation. You may also specify --dbpath (page 1023) on the command line; however, always prefer the conguration le. If the dbpath (page 1081) directory does not exist, mongod.exe (page 1041) will not be able to start. The default value for dbpath (page 1081) is \data\db.
Stop or Remove the MongoDB Service To stop the MongoDB service:
1.6. Install MongoDB on Windows
19
net stop MongoDB
To remove the MongoDB service:

C:\mongodb\bin\mongod.exe --remove
1.7 Install MongoDB Enterprise

New in version 2.2. MongoDB Enterprise is available on four platforms and contains support for several features related to security and monitoring.
1.7.1 Required Packages

Changed in version 2.4: MongoDB Enterprise requires libgsasl. To use MongoDB Enterprise, you must install several prerequisites. The names of the packages vary by distribution and are as follows: Ubuntu 12.04 and 11.04 require libssl0.9.8, libgsasl, snmp, and snmpd. Issue a command such as the following to install these packages:
sudo apt-get install libssl0.9.8 libgsasl7 snmp snmpd
Red Hat Enterprise Linux 6.x series and Amazon Linux AMI require libssl, libgsasl7, net-snmp, net-snmp-libs, and net-snmp-utils. To download libgsasl you must enable the EPEL repository by issuing the following sequence of commands to add and update the system repositories:
sudo rpm -ivh http://download.fedoraproject.org/pub/epel/6/x86_64/epel-release-6-8.noarch.rpm sudo yum update -y
When you have installed and updated the EPEL repositories, issue the following install these packages:
sudo yum install libssl net-snmp net-snmp-libs net-snmp-utils libgsasl
SUSE Enterprise Linux requires libopenssl0_9_8, libsnmp15, slessp1-libsnmp15, and snmp-mibs. Issue a command such as the following to install these packages:
sudo zypper install libopenssl0_9_8 libsnmp15 slessp1-libsnmp15 snmp-mibs
Note: For the 2.4 release, the MongoDB Enterprise for SUSE requires libgsasl which is not available in the default repositories for SUSE.
1.7.2 Install MongoDB Enterprise Binaries

When you have installed the required packages, and downloaded the Enterprise packages you can install the packages using the same procedure as a standard installation of MongoDB on Linux Systems (page 11). After you have installed MongoDB, consider the following documents as you begin to learn about MongoDB:
20
1.8 Getting Started with MongoDB

This tutorial provides an introduction to basic database operations using the mongo (page 1036) shell. mongo (page 1036) is a part of the standard MongoDB distribution and provides a full JavaScript environment with a complete access to the JavaScript language and all standard functions as well as a full database interface for MongoDB. See the mongo JavaScript API documentation and the mongo (page 1036) shell JavaScript Method Reference (page 917). The tutorial assumes that youre running MongoDB on a Linux or OS X operating system and that you have a running database server; MongoDB does support Windows and provides a Windows distribution with identical operation. For instructions on installing MongoDB and starting the database server see the appropriate installation (page 3) document. This tutorial addresses the following aspects of MongoDB use: Connect to a Database (page 21) Connect to a mongod (page 1021) (page 21) Select a Database (page 21) Display mongo Help (page 22) Create a Collection and Insert Documents (page 22) Insert Individual Documents (page 22) Insert Multiple Documents Using a For Loop (page 23) Working with the Cursor (page 24) Iterate over the Cursor with a Loop (page 24) Use Array Operations with the Cursor (page 25) Query for Specic Documents (page 25) Return a Single Document from a Collection (page 26) Limit the Number of Documents in the Result Set (page 27) Next Steps with MongoDB (page 27)
1.8.1 Connect to a Database

In this section you connect to the database server, which runs as mongod (page 1021), and begin using the mongo (page 1036) shell to select a logical database within the database instance and access the help text in the mongo (page 1036) shell. Connect to a mongod From a system prompt, start mongo (page 1036) by issuing the mongo (page 1036) command, as follows:
mongo
By default, mongo (page 1036) looks for a database server listening on port 27017 on the localhost interface. To connect to a server on a different port or interface, use the --port (page 1037) and --host (page 1037) options. Select a Database After starting the mongo (page 1036) shell your session will use the test database for context, by default. At any time issue the following operation at the mongo (page 1036) to report the current database:
db
db returns the name of the current database.
1.8. Getting Started with MongoDB
21
1. From the mongo (page 1036) shell, display the list of databases with the following operation:
show dbs
2. Switch to a new database named mydb with the following operation:

use mydb
3. Conrm that your session has the mydb database as context, using the db operation, which returns the name of the current database as follows:
db
At this point, if you issue the show dbs operation again, it will not include mydb, because MongoDB will not create a database until you insert data into that database. The Create a Collection and Insert Documents (page 22) section describes the process for inserting data. New in version 2.4: show databases also returns a list of databases. Display mongo Help At any point you can access help for the mongo (page 1036) shell using the following operation:
help
Furthermore, you can append the .help() method to some JavaScript methods, any cursor object, as well as the db and db.collection objects to return additional help information.
1.8.2 Create a Collection and Insert Documents

In this section, you insert documents into a new collection named things within the new database named mydb. MongoDB will create collections and databases implicitly upon their rst use: you do not need to create the database or collection before inserting data. Furthermore, because MongoDB uses dynamic schemas (page 710), you do not need to specify the structure of your documents before inserting them into the collection. Insert Individual Documents 1. From the mongo (page 1036) shell, conrm that the current context is the mydb database with the following operation:
db
2. If mongo (page 1036) does not return mydb for the previous operation, set the context to the mydb database with the following operation:
use mydb
3. Create two documents, named j and k, with the following sequence of JavaScript operations:
j = { name : "mongo" } k = { x : 3 }
4. Insert the j and k documents into the collection things with the following sequence of operations:
db.things.insert( j ) db.things.insert( k )
22
When you insert the rst document, the mongod (page 1021) will create both the mydb database and the things collection. 5. Conrm that the collection named things exists using the following operation:
show collections
The mongo (page 1036) shell will return the list of the collections in the current (i.e. mydb) database. At this point, the only collection is things. All mongod (page 1021) databases also have a system.indexes (page 1099) collection. 6. Conrm that the documents exist in the collection things by issuing query on the things collection. Using the find() (page 924) method in an operation that resembles the following:
db.things.find()
This operation returns the following results. The ObjectId (page 194) values will be unique:
{ "_id" : ObjectId("4c2209f9f3924d31102bd84a"), "name" : "mongo" } { "_id" : ObjectId("4c2209fef3924d31102bd84b"), "x" : 3 }
All MongoDB documents must have an _id eld with a unique value. These operations do not explicitly specify a value for the _id eld, so mongo (page 1036) creates a unique ObjectId (page 194) value for the eld before inserting it into the collection. Insert Multiple Documents Using a For Loop 1. From the mongo (page 1036) shell, add more documents to the things collection using the following for loop:
for (var i = 1; i <= 20; i++) db.things.insert( { x : 4 , j : i } )
2. Query the collection by issuing the following command:

db.things.find()
The mongo (page 1036) shell displays the rst 20 documents in the collection. Your ObjectId (page 194) values will be different:
{ { { { { { { { { { { { { { { { { { { { "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" : : : : : : : : : : : : : : : : : : : : ObjectId("4c2209f9f3924d31102bd84a"), ObjectId("4c2209fef3924d31102bd84b"), ObjectId("4c220a42f3924d31102bd856"), ObjectId("4c220a42f3924d31102bd857"), ObjectId("4c220a42f3924d31102bd858"), ObjectId("4c220a42f3924d31102bd859"), ObjectId("4c220a42f3924d31102bd85a"), ObjectId("4c220a42f3924d31102bd85b"), ObjectId("4c220a42f3924d31102bd85c"), ObjectId("4c220a42f3924d31102bd85d"), ObjectId("4c220a42f3924d31102bd85e"), ObjectId("4c220a42f3924d31102bd85f"), ObjectId("4c220a42f3924d31102bd860"), ObjectId("4c220a42f3924d31102bd861"), ObjectId("4c220a42f3924d31102bd862"), ObjectId("4c220a42f3924d31102bd863"), ObjectId("4c220a42f3924d31102bd864"), ObjectId("4c220a42f3924d31102bd865"), ObjectId("4c220a42f3924d31102bd866"), ObjectId("4c220a42f3924d31102bd867"), "name" : "mongo" } "x" : 3 } "x" : 4, "j" : 1 } "x" : 4, "j" : 2 } "x" : 4, "j" : 3 } "x" : 4, "j" : 4 } "x" : 4, "j" : 5 } "x" : 4, "j" : 6 } "x" : 4, "j" : 7 } "x" : 4, "j" : 8 } "x" : 4, "j" : 9 } "x" : 4, "j" : 10 } "x" : 4, "j" : 11 } "x" : 4, "j" : 12 } "x" : 4, "j" : 13 } "x" : 4, "j" : 14 } "x" : 4, "j" : 15 } "x" : 4, "j" : 16 } "x" : 4, "j" : 17 } "x" : 4, "j" : 18 }
23
1. The find() (page 924) returns a cursor. To iterate the cursor and return more documents use the it operation in the mongo (page 1036) shell. The mongo (page 1036) shell will exhaust the cursor, and return the following documents:
{ "_id" : ObjectId("4c220a42f3924d31102bd868"), "x" : 4, "j" : 19 } { "_id" : ObjectId("4c220a42f3924d31102bd869"), "x" : 4, "j" : 20 }
For more information on inserting new documents, see the insert() (page 202) documentation.
1.8.3 Working with the Cursor

When you query a collection, MongoDB returns a cursor object that contains the results of the query. The mongo (page 1036) shell then iterates over the cursor to display the results. Rather than returning all results at once, the shell iterates over the cursor 20 times to display the rst 20 results and then waits for a request to iterate over the remaining results. This prevents mongo (page 1036) from displaying thousands or millions of results at once. The it operation allows you to iterate over the next 20 results in the shell. In the previous procedure (page 23), the cursor only contained two more documents, and so only two more documents displayed. The procedures in this section show other ways to work with a cursor. For comprehensive documentation on cursors, see Iterate the Returned Cursor (page 216). Iterate over the Cursor with a Loop 1. In the MongoDB JavaScript shell, query the things collection and assign the resulting cursor object to the c variable:
var c = db.things.find()
2. Print the full result set by using a while loop to iterate over the c variable:
while ( c.hasNext() ) printjson( c.next() )
The hasNext() function returns true if the cursor has documents. The next() method returns the next document. The printjson() method renders the document in a JSON-like format. The result of this operation follows, although if the ObjectId (page 194) values will be unique:
{ { { { { { { { { { { { { { { { { { { { "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" : : : : : : : : : : : : : : : : : : : : ObjectId("4c2209f9f3924d31102bd84a"), ObjectId("4c2209fef3924d31102bd84b"), ObjectId("4c220a42f3924d31102bd856"), ObjectId("4c220a42f3924d31102bd857"), ObjectId("4c220a42f3924d31102bd858"), ObjectId("4c220a42f3924d31102bd859"), ObjectId("4c220a42f3924d31102bd85a"), ObjectId("4c220a42f3924d31102bd85b"), ObjectId("4c220a42f3924d31102bd85c"), ObjectId("4c220a42f3924d31102bd85d"), ObjectId("4c220a42f3924d31102bd85e"), ObjectId("4c220a42f3924d31102bd85f"), ObjectId("4c220a42f3924d31102bd860"), ObjectId("4c220a42f3924d31102bd861"), ObjectId("4c220a42f3924d31102bd862"), ObjectId("4c220a42f3924d31102bd863"), ObjectId("4c220a42f3924d31102bd864"), ObjectId("4c220a42f3924d31102bd865"), ObjectId("4c220a42f3924d31102bd866"), ObjectId("4c220a42f3924d31102bd867"), "name" : "mongo" } "x" : 3 } "x" : 4, "j" : 1 } "x" : 4, "j" : 2 } "x" : 4, "j" : 3 } "x" : 4, "j" : 4 } "x" : 4, "j" : 5 } "x" : 4, "j" : 6 } "x" : 4, "j" : 7 } "x" : 4, "j" : 8 } "x" : 4, "j" : 9 } "x" : 4, "j" : 10 } "x" : 4, "j" : 11 } "x" : 4, "j" : 12 } "x" : 4, "j" : 13 } "x" : 4, "j" : 14 } "x" : 4, "j" : 15 } "x" : 4, "j" : 16 } "x" : 4, "j" : 17 } "x" : 4, "j" : 18 }
24
{ "_id" : ObjectId("4c220a42f3924d31102bd868"), "x" : 4, "j" : 19 } { "_id" : ObjectId("4c220a42f3924d31102bd869"), "x" : 4, "j" : 20 }
Use Array Operations with the Cursor You can manipulate a cursor object as if it were an array. Consider the following procedure: 1. In the mongo (page 1036) shell, query the things collection and assign the resulting cursor object to the c variable:
var c = db.things.find()
2. To nd the document at the array index 4, use the following operation:

printjson( c [ 4 ] )
MongoDB returns the following:

{ "_id" : ObjectId("4c220a42f3924d31102bd858"), "x" : 4, "j" : 3 }
When you access documents in a cursor using the array index notation, mongo (page 1036) rst calls the cursor.toArray() method and loads into RAM all documents returned by the cursor. The index is then applied to the resulting array. This operation iterates the cursor completely and exhausts the cursor. For very large result sets, mongo (page 1036) may run out of available memory. For more information on the cursor, see Iterate the Returned Cursor (page 216). Query for Specic Documents MongoDB has a rich query system that allows you to select and lter the documents in a collection along specic elds and values. See Query Document (page 168) and Read (page 209) for a full account of queries in MongoDB. In this procedure, you query for specic documents in the things collection by passing a query document as a parameter to the find() (page 924) method. A query document species the criteria the query must match to return a document. To query for specic documents, do the following: 1. In the mongo (page 1036) shell, query for all documents where the name eld has a value of mongo by passing the { name : "mongo" } query document as a parameter to the find() (page 924) method:
db.things.find( { name : "mongo" } )
MongoDB returns one document that ts this criteria. The ObjectId (page 194) value will be different:
{ "_id" : ObjectId("4c2209f9f3924d31102bd84a"), "name" : "mongo" }
2. Query for all documents where x has a value of 4 by passing the { x : to find() (page 924):
db.things.find( { x : 4 } )
4 } query document as a parameter
MongoDB returns the following result set:

{ { { { "_id" "_id" "_id" "_id" : : : : ObjectId("4c220a42f3924d31102bd856"), ObjectId("4c220a42f3924d31102bd857"), ObjectId("4c220a42f3924d31102bd858"), ObjectId("4c220a42f3924d31102bd859"), "x" "x" "x" "x" : : : : 4, 4, 4, 4, "j" "j" "j" "j" : : : : 1 2 3 4 } } } }
25
{ { { { { { { { { { { { { { { {
"_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id"
: : : : : : : : : : : : : : : :
ObjectId("4c220a42f3924d31102bd85a"), ObjectId("4c220a42f3924d31102bd85b"), ObjectId("4c220a42f3924d31102bd85c"), ObjectId("4c220a42f3924d31102bd85d"), ObjectId("4c220a42f3924d31102bd85e"), ObjectId("4c220a42f3924d31102bd85f"), ObjectId("4c220a42f3924d31102bd860"), ObjectId("4c220a42f3924d31102bd861"), ObjectId("4c220a42f3924d31102bd862"), ObjectId("4c220a42f3924d31102bd863"), ObjectId("4c220a42f3924d31102bd864"), ObjectId("4c220a42f3924d31102bd865"), ObjectId("4c220a42f3924d31102bd866"), ObjectId("4c220a42f3924d31102bd867"), ObjectId("4c220a42f3924d31102bd868"), ObjectId("4c220a42f3924d31102bd869"),
"x" "x" "x" "x" "x" "x" "x" "x" "x" "x" "x" "x" "x" "x" "x" "x"
: : : : : : : : : : : : : : : :
4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
"j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j"
: : : : : : : : : : : : : : : :
5 } 6 } 7 } 8 } 9 } 10 } 11 } 12 } 13 } 14 } 15 } 16 } 17 } 18 } 19 } 20 }
ObjectId (page 194) values are always unique. 3. Query for all documents where x has a value of 4, as in the previous query, but only return only the value of j. MongoDB will also return the _id eld, unless explicitly excluded. To do this, you add the { j : 1 } document as the projection in the second parameter to find() (page 924). This operation would resemble the following:
db.things.find( { x : 4 } , { j : 1 } )
MongoDB returns the following results:

{ { { { { { { { { { { { { { { { { { { { "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" : : : : : : : : : : : : : : : : : : : : ObjectId("4c220a42f3924d31102bd856"), ObjectId("4c220a42f3924d31102bd857"), ObjectId("4c220a42f3924d31102bd858"), ObjectId("4c220a42f3924d31102bd859"), ObjectId("4c220a42f3924d31102bd85a"), ObjectId("4c220a42f3924d31102bd85b"), ObjectId("4c220a42f3924d31102bd85c"), ObjectId("4c220a42f3924d31102bd85d"), ObjectId("4c220a42f3924d31102bd85e"), ObjectId("4c220a42f3924d31102bd85f"), ObjectId("4c220a42f3924d31102bd860"), ObjectId("4c220a42f3924d31102bd861"), ObjectId("4c220a42f3924d31102bd862"), ObjectId("4c220a42f3924d31102bd863"), ObjectId("4c220a42f3924d31102bd864"), ObjectId("4c220a42f3924d31102bd865"), ObjectId("4c220a42f3924d31102bd866"), ObjectId("4c220a42f3924d31102bd867"), ObjectId("4c220a42f3924d31102bd868"), ObjectId("4c220a42f3924d31102bd869"), "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" "j" : : : : : : : : : : : : : : : : : : : : 1 } 2 } 3 } 4 } 5 } 6 } 7 } 8 } 9 } 10 } 11 } 12 } 13 } 14 } 15 } 16 } 17 } 18 } 19 } 20 }
Return a Single Document from a Collection With the db.collection.findOne() (page 929) method you can return a single document from a MongoDB collection. The findOne() (page 929) method takes the same parameters as find() (page 924), but returns a document rather than a cursor. To retrieve one document from the things collection, issue the following command:
26
db.things.findOne()
For more information on querying for documents, see the Read (page 209) and Read Operations (page 167) documentation. Limit the Number of Documents in the Result Set You can constrain the size of the result set to increase performance by limiting the amount of data your application must receive over the network. To specify the maximum number of documents in the result set, call the limit() (page 959) method on a cursor, as in the following command:
db.things.find().limit(3)
MongoDB will return the following result, with different ObjectId (page 194) values:
{ "_id" : ObjectId("4c2209f9f3924d31102bd84a"), "name" : "mongo" } { "_id" : ObjectId("4c2209fef3924d31102bd84b"), "x" : 3 } { "_id" : ObjectId("4c220a42f3924d31102bd856"), "x" : 4, "j" : 1 }
1.8.4 Next Steps with MongoDB

For more information on manipulating the documents in a database as you continue to learn MongoDB, consider the following resources: CRUD Operations for MongoDB (page 201) SQL to MongoDB Mapping Chart (page 1016) MongoDB Drivers and Client Libraries (page 555) Getting Started with MongoDB (page 21) Create (page 201) Read (page 209) Update (page 219) Delete (page 227)
27
28
CHAPTER 2
Release Notes
You should always install the latest, stable version of MongoDB. Stable versions have an even-numbered minor version number. For example: v2.4 is stable, v2.2, and v2.0 were previously the stable, while v2.1 and v2.3 are a development versions. Current Stable Release: Release Notes for MongoDB 2.4 (page 1135) Previous Stable Releases: Release Notes for MongoDB 2.2 (page 1157) Release Notes for MongoDB 2.0 (page 1167) Release Notes for MongoDB 1.8 (page 1173)
29
30
Chapter 2. Release Notes
Part II
Administration
31
The documentation in this section outlines core administrative tasks and practices that operators of MongoDB will want to consider.
33
34
CHAPTER 3
Run-time Database Conguration
The command line (page 1021) and conguration le (page 1078) interfaces provide MongoDB administrators with a large number of options and settings for controlling the operation of the database system. This document provides an overview of common congurations and examples of best-practice congurations for common use cases. While both interfaces provide access to the same collection of options and settings, this document primarily uses the conguration le interface. If you run MongoDB using a control script or installed from a package for your operating system, you likely already have a conguration le located at http://docs.mongodb.org/manual/etc/mongodb.conf. Conrm this by checking the content of the http://docs.mongodb.org/manual/etc/init.d/mongod or http://docs.mongodb.org/manual/etc/rc.d/mongod script to insure that the control scripts start the mongod (page 1021) with the appropriate conguration le (see below.) To start MongoDB instance using this conguration issue a command in the following form:
mongod --config /etc/mongodb.conf mongod -f /etc/mongodb.conf
Modify the values in the http://docs.mongodb.org/manual/etc/mongodb.conf le on your system to control the conguration of your database instance.
3.1 Congure the Database

Consider the following basic conguration:
fork = true bind_ip = 127.0.0.1 port = 27017 quiet = true dbpath = /srv/mongodb logpath = /var/log/mongodb/mongod.log logappend = true journal = true
For most standalone servers, this is a sufcient base conguration. It makes several assumptions, but consider the following explanation: fork (page 1081) is true, which enables a daemon mode for mongod (page 1021), which detaches (i.e. forks) the MongoDB from the current session and allows you to run the database as a conventional server. bind_ip (page 1079) is 127.0.0.1, which forces the server to only listen for requests on the localhost IP. Only bind to secure interfaces that the application-level systems can access with access control provided by system network ltering (i.e. rewall). 35
port (page 1079) is 27017, which is the default MongoDB port for database instances. MongoDB can bind to any port. You can also lter access based on port using network ltering tools. Note: UNIX-like systems require superuser privileges to attach processes to ports lower than 1024. quiet (page 1086) is true. This disables all but the most critical entries in output/log le. In normal operation this is the preferable operation to avoid log noise. In diagnostic or testing situations, set this value to false. Use setParameter (page 869) to modify this setting during run time. dbpath (page 1081) is http://docs.mongodb.org/manual/srv/mongodb, which species where MongoDB will store its data les. http://docs.mongodb.org/manual/srv/mongodb and http://docs.mongodb.org/manual/var/lib/mongodb are popular locations. The user account that mongod (page 1021) runs under will need read and write access to this directory. logpath (page 1080) is http://docs.mongodb.org/manual/var/log/mongodb/mongod.log which is where mongod (page 1021) will write its output. If you do not set this value, mongod (page 1021) writes all output to standard output (e.g. stdout.) logappend (page 1080) is true, which ensures that mongod (page 1021) does not overwrite an existing log le following the server start operation. journal (page 1083) is true, which enables journaling. Journaling ensures single instance write-durability. 64-bit builds of mongod (page 1021) enable journaling by default. Thus, this setting may be redundant. Given the default conguration, some of these values may be redundant. However, in many situations explicitly stating the conguration increases overall system intelligibility.
3.2 Security Considerations

The following collection of conguration options are useful for limiting access to a mongod (page 1021) instance. Consider the following:
bind_ip = 127.0.0.1,10.8.0.10,192.168.4.24 nounixsocket = true auth = true
Consider the following explanation for these conguration decisions: bind_ip (page 1079) has three values: 127.0.0.1, the localhost interface; 10.8.0.10, a private IP address typically used for local networks and VPN interfaces; and 192.168.4.24, a private network interface typically used for local networks. Because production MongoDB instances need to be accessible from multiple database servers, it is important to bind MongoDB to multiple interfaces that are accessible from your application servers. At the same time its important to limit these interfaces to interfaces controlled and protected at the network layer. nounixsocket (page 1081) to true disables the UNIX Socket, which is otherwise enabled by default. This limits access on the local system. This is desirable when running MongoDB on systems with shared access, but in most situations has minimal impact. auth (page 1081) is true enables the authentication system within MongoDB. If enabled you will need to log in by connecting over the localhost interface for the rst time to create user credentials. See also: Security Practices and Management (page 131)
36
Chapter 3. Run-time Database Conguration
3.3 Replication and Sharding Conguration

3.3.1 Replication Conguration
Replica set conguration is straightforward, and only requires that the replSet (page 1088) have a value that is consistent among all members of the set. Consider the following:
replSet = set0
Use descriptive names for sets. Once congured use the mongo (page 1036) shell to add hosts to the replica set. See also: Replica set reconguration (page 467). To enable authentication for the replica set, add the following option:
keyFile = /srv/mongodb/keyfile
New in version 1.8: for replica sets, and 1.9.1 for sharded replica sets. Setting keyFile (page 1081) enables authentication and species a key le for the replica set member use to when authenticating to each other. The content of the key le is arbitrary, but must be the same on all members of the replica set and mongos (page 1032) instances that connect to the set. The keyle must be less than one kilobyte in size and may only contain characters in the base64 set and the le must not have group or world permissions on UNIX systems. See also: The Replica set Reconguration (page 467) section for information regarding the process for changing replica set during operation. Additionally, consider the Replica Set Security (page 393) section for information on conguring authentication with replica sets. Finally, see the Replication (page 385) index and the Replica Set Fundamental Concepts (page 387) document for more information on replication in MongoDB and replica set conguration in general.
3.3.2 Sharding Conguration

Sharding requires a number of mongod (page 1021) instances with different congurations. The cong servers store the clusters metadata, while the cluster distributes data among one or more shard servers. Note: Cong servers are not replica sets. To set up one or three cong server instances as normal (page 35) mongod (page 1021) instances, and then add the following conguration option:
configsvr = true bind_ip = 10.8.0.12 port = 27001
This creates a cong server running on the private IP address 10.8.0.12 on port 27001. Make sure that there are no port conicts, and that your cong server is accessible from all of your mongos (page 1032) and mongod (page 1021) instances. To set up shards, congure two or more mongod (page 1021) instance using your base conguration (page 35), adding the shardsvr (page 1089) setting: 3.3. Replication and Sharding Conguration 37
shardsvr = true
Finally, to establish the cluster, congure at least one mongos (page 1032) process with the following settings:
configdb = 10.8.0.12:27001 chunkSize = 64
You can specify multiple configdb (page 1089) instances by specifying hostnames and ports in the form of a comma separated list. In general, avoid modifying the chunkSize (page 1090) from the default value of 64, 1 and should ensure this setting is consistent among all mongos (page 1032) instances. See also: The Sharding (page 483) section of the manual for more information on sharding and cluster conguration.
3.4 Run Multiple Database Instances on the Same System

In many cases running multiple instances of mongod (page 1021) on a single system is not recommended. On some types of deployments 2 and for testing purposes you may need to run more than one mongod (page 1021) on a single system. In these cases, use a base conguration (page 35) for each instance, but consider the following conguration values:
dbpath = /srv/mongodb/db0/ pidfilepath = /srv/mongodb/db0.pid
The dbpath (page 1081) value controls the location of the mongod (page 1021) instances data directory. Ensure that each database has a distinct and well labeled data directory. The pidfilepath (page 1081) controls where mongod (page 1021) process places its process id le. As this tracks the specic mongod (page 1021) le, it is crucial that le be unique and well labeled to make it easy to start and stop these processes. Create additional control scripts and/or adjust your existing MongoDB conguration and control script as needed to control these processes.
3.5 Diagnostic Congurations

The following conguration options control various mongod (page 1021) behaviors for diagnostic purposes. The following settings have default values that tuned for general production purposes:
slowms = 50 profile = 3 verbose = true diaglog = 3 objcheck = true cpu = true
Use the base conguration (page 35) and add these options if you are experiencing some unknown issue or performance problem as needed: slowms (page 1085) congures the threshold for the database proler to consider a query slow. The default value is 100 milliseconds. Set a lower value if the database proler does not return useful results. See Optimization Strategies for MongoDB (page 555) for more information on optimizing operations in MongoDB.
1 Chunk size is 64 megabytes by default, which provides the ideal balance between the most even distribution of data, for which smaller chunk sizes are best, and minimizing chunk migration, for which larger chunk sizes are optimal. 2 Single-tenant systems with SSD or other high performance disks may provide acceptable performance levels for multiple mongod (page 1021) instances. Additionally, you may nd that multiple databases with small working sets may function acceptably on a single system.
38
profile (page 1084) sets the database proler level. The proler is not active by default because of the possible impact on the proler itself on performance. Unless this setting has a value, queries are not proled. verbose (page 1079) enables a verbose logging mode that modies mongod (page 1021) output and increases logging to include a greater number of events. Only use this option if you are experiencing an issue that is not reected in the normal logging level. If you require additional verbosity, consider the following options:
v = true vv = true vvv = true vvvv = true vvvvv = true
Each additional level v adds additional verbosity to the logging. The verbose option is equal to v = true. diaglog (page 1082) enables diagnostic logging. Level 3 logs all read and write options. objcheck (page 1080) forces mongod (page 1021) to validate all requests from clients upon receipt. Use this option to ensure that invalid requests are not causing errors, particularly when running a database with untrusted clients. This option may affect database performance. cpu (page 1081) forces mongod (page 1021) to report the percentage of the last interval spent in write-lock. The interval is typically 4 seconds, and each output line in the log includes both the actual interval since the last report and the percentage of time spent in write lock.
3.5. Diagnostic Congurations
39
40
CHAPTER 4
Backup and Recovery Operations for MongoDB
4.1 Backup Strategies for MongoDB Systems

Backups are an important part of any operational disaster recovery plan. A good backup plan must be able to capture data in a consistent and usable state, and operators must be able to automate both the backup and the recovery operations. Also test all components of the backup system to ensure that you can recover backed up data as needed. If you cannot effectively restore your database from the backup, then your backups are useless. This document addresses higher level backup strategies, for more information on specic backup procedures consider the following documents: Use Filesystem Snapshots to Backup and Restore MongoDB Databases (page 46). Use mongodump and mongorestore to Backup and Restore MongoDB Databases (page 43). Backup a Small Sharded Cluster with mongodump (page 54) Create Backup of a Sharded Cluster with Filesystem Snapshots (page 55) Create Backup of a Sharded Cluster with Database Dumps (page 56) Schedule Backup Window for Sharded Clusters (page 59) Restore a Single Shard (page 58) Restore Sharded Clusters (page 58)
4.1.1 Backup Considerations

As you develop a backup strategy for your MongoDB deployment consider the following factors: Geography. Ensure that you move some backups away from the your primary database infrastructure. System errors. Ensure that your backups can survive situations where hardware failures or disk errors impact the integrity or availability of your backups. Production constraints. Backup operations themselves sometimes require substantial system resources. It is important to consider the time of the backup schedule relative to peak usage and maintenance windows. System capabilities. Some of the block-level snapshot tools require special support on the operating-system or infrastructure level. Database conguration. Replication and sharding can affect the process and impact of the backup implementation. See Sharded Cluster Backup Considerations (page 42) and Replica Set Backup Considerations (page 43). Actual requirements. You may be able to save time, effort, and space by including only crucial data in the most frequent backups and backing up less crucial data less frequently.
41
4.1.2 Approaches to Backing Up MongoDB Systems

There are two main methodologies for backing up MongoDB instances. Creating binary dumps of the database using mongodump (page 1044) or creating lesystem level snapshots. Both methodologies have advantages and disadvantages: binary database dumps are comparatively small, because they dont include index content or pre-allocated free space, and record padding (page 181). However, its impossible to capture a copy of a running system that reects a single moment in time using a binary dump. lesystem snapshots, sometimes called block level backups, produce larger backup sizes, but complete quickly and can reect a single moment in time on a running system. However, snapshot systems require lesystem and operating system support and tools. The best option depends on the requirements of your deployment and disaster recovery needs. Typically, lesystem snapshots are because of their accuracy and simplicity; however, mongodump (page 1044) is a viable option used often to generate backups of MongoDB systems. The following documents provide details and procedures on the two approaches: Use Filesystem Snapshots to Backup and Restore MongoDB Databases (page 46). Use mongodump and mongorestore to Backup and Restore MongoDB Databases (page 43). In some cases, taking backups is difcult or impossible because of large data volumes, distributed architectures, and data transmission speeds. In these situations, increase the number of members in your replica set or sets.
4.1.3 Backup Strategies for MongoDB Deployments

Sharded Cluster Backup Considerations Important: To capture a point-in-time backup from a sharded cluster you must stop all writes to the cluster. On a running production system, you can only capture an approximation of point-in-time snapshot. Sharded clusters complicate backup operations, as distributed systems. True point-in-time backups are only possible when stopping all write activity from the application. To create a precise moment-in-time snapshot of a cluster, stop all application write activity to the database, capture a backup, and allow only write operations to the database after the backup is complete. However, you can capture a backup of a cluster that approximates a point-in-time backup by capturing a backup from a secondary member of the replica sets that provide the shards in the cluster at roughly the same moment. If you decide to use an approximate-point-in-time backup method, ensure that your application can operate using a copy of the data that does not reect a single moment in time. The following documents describe sharded cluster related backup procedures: Backup a Small Sharded Cluster with mongodump (page 54) Create Backup of a Sharded Cluster with Filesystem Snapshots (page 55) Create Backup of a Sharded Cluster with Database Dumps (page 56) Schedule Backup Window for Sharded Clusters (page 59) Restore a Single Shard (page 58) Restore Sharded Clusters (page 58)
42
Chapter 4. Backup and Recovery Operations for MongoDB
Replica Set Backup Considerations In most cases, backing up data stored in a replica set is similar to backing up data stored in a single instance. It is possible to lock a single secondary database and then create a backup from that instance. When you unlock the database, the secondary will catch up with the primary. You may also choose to deploy a dedicated hidden member for backup purposes. If you have a sharded cluster where each shard is itself a replica set, you can use this method to create a backup of the entire cluster without disrupting the operation of the node. In these situations you should still turn off the balancer when you create backups. For any cluster, using a non-primary node to create backups is particularly advantageous in that the backup operation does not affect the performance of the primary. Replication itself provides some measure of redundancy. Nevertheless, keeping point-in time backups of your cluster to provide for disaster recovery and as an additional layer of protection is crucial. For an overview of backup strategies and considerations for all MongoDB deployments, consider, Backup Strategies for MongoDB Systems (page 41). For practical instructions and example backup procedures consider the following documents:
4.2 Backup and Recovery Procedures

4.2.1 Use mongodump and mongorestore to Backup and Restore MongoDB Databases
This document describes the process for writing the entire contents of your MongoDB instance to a le in a binary format. If disk-level snapshots are not available, this approach provides the best option for full system database backups. If your system has disk level snapshot capabilities, consider the backup methods described in Use Filesystem Snapshots to Backup and Restore MongoDB Databases (page 46). See also: Backup Strategies for MongoDB Systems (page 41) mongodump (page 1044) mongorestore (page 1047) Backup a Database with mongodump
Basic mongodump Operations
The mongodump (page 1044) utility can back up data by either: connecting to a running mongod (page 1021) or mongos (page 1032) instance, or accessing data les without an active instance. The utility can create a backup for an entire server, database or collection, or can use a query to backup just part of a collection. When you run mongodump (page 1044) without any arguments, the command connects to the local database instance (e.g. 127.0.0.1 or localhost) on port 27017 and creates a database backup named dump/ in the current directory. To backup data from a mongod (page 1021) or mongos (page 1032) instance running on the same machine and on the default port of 27017 use the following command: 4.2. Backup and Recovery Procedures 43
mongodump
Note: The format of data created by mongodump (page 1044) tool from the 2.2 distribution or later is different and incompatible with earlier versions of mongod (page 1021). To limit the amount of data included in the database dump, you can specify --db (page 1045) and --collection (page 1046) as options to the mongodump (page 1044) command. For example:
mongodump --dbpath /data/db/ --out /data/backup/ mongodump --host mongodb.example.net --port 27017
mongodump (page 1044) will write BSON les that hold a copy of data accessible via the mongod (page 1021) listening on port 27017 of the mongodb.example.net host.
mongodump --collection collection --db test
This command creates a dump of the collection named collection from the database test in a dump/ subdirectory of the current working directory.
Point in Time Operation Using Oplogs
Use the --oplog (page 1046) option with mongodump (page 1044) to collect the oplog entries to build a point-intime snapshot of a database within a replica set. With --oplog (page 1046), mongodump (page 1044) copies all the data from the source database as well as all of the oplog entries from the beginning of the backup procedure to until the backup procedure completes. This backup procedure, in conjunction with mongorestore --oplogReplay (page 1050), allows you to restore a backup that reects a consistent and specic moment in time.
Create Backups Without a Running mongod Instance
If your MongoDB instance is not running, you can use the --dbpath (page 1045) option to specify the location to your MongoDB instances database les. mongodump (page 1044) reads from the data les directly with this operation. This locks the data directory to prevent conicting writes. The mongod (page 1021) process must not be running or attached to these data les when you run mongodump (page 1044) in this conguration. Consider the following example:
mongodump --dbpath /srv/mongodb
Create Backups from Non-Local mongod Instances
The --host (page 1044) and --port (page 1044) options for mongodump (page 1044) allow you to connect to and backup from a remote host. Consider the following example:
mongodump --host mongodb1.example.net --port 3017 --username user --password pass --out /opt/backup/m
On any mongodump (page 1044) command you may, as above, specify username and password credentials to specify database authentication. Restore a Database with mongorestore The mongorestore (page 1048) utility restores a binary backup created by mongodump (page 1044). By default, mongorestore (page 1048) looks for a database backup in the dump/ directory. 44 Chapter 4. Backup and Recovery Operations for MongoDB
The mongorestore (page 1048) utility can restore data either by: connecting to a running mongod (page 1021) or mongos (page 1032) directly, or writing to a local database path without use of a running mongod (page 1021). The mongorestore (page 1048) utility can restore either an entire database backup or a subset of the backup. A mongorestore (page 1048) command that connects to an active mongod (page 1021) or mongos (page 1032) has the following prototype form:
mongorestore --port <port number> <path to the backup>
A mongorestore (page 1048) command that writes to data les without using a running mongod (page 1021) has the following prototype form:
mongorestore --dbpath <local database path> <path to the backup>
Consider the following example:

mongorestore dump-2012-10-25/
Here, mongorestore (page 1048) imports the database backup in the dump-2012-10-25 directory to the mongod (page 1021) instance running on the localhost interface.
Restore Point in Time Oplog Backup
If you created your database dump using the --oplog (page 1046) option to ensure a point-in-time snapshot, call mongorestore (page 1048) with the --oplogReplay (page 1050) option, as in the following example:
mongorestore --oplogReplay
You may also consider using the mongorestore --objcheck (page 1050) option to check the integrity of objects while inserting them into the database, or you may consider the mongorestore --drop (page 1050) option to drop each collection from the database before restoring from backups.
Restore a Subset of data from a Binary Database Dump
mongorestore (page 1048) also includes the ability to a lter to all input before inserting it into the new database. Consider the following example:
mongorestore --filter {"field": 1}
Here, mongorestore (page 1048) only adds documents to the database from the dump located in the dump/ folder if the documents have a eld name field that holds a value of 1. Enclose the lter in single quotes (e.g. ) to prevent the lter from interacting with your shell environment.
Restore without a Running mongod
mongorestore (page 1048) can write data to MongoDB data les without needing to connect to a mongod (page 1021) directly.
mongorestore --dbpath /srv/mongodb --journal
Here, mongorestore (page 1048) restores the database dump located in dump/ folder into the data les located at http://docs.mongodb.org/manual/srv/mongodb. Additionally, the --journal (page 1049) option
4.2. Backup and Recovery Procedures
45
ensures that mongorestore (page 1048) records all operation in the durability journal. The journal prevents data le corruption if anything (e.g. power failure, disk failure, etc.) interrupts the restore operation. See also: mongodump (page 1044) and mongorestore (page 1047).
Restore Backups to Non-Local mongod Instances
By default, mongorestore (page 1048) connects to a MongoDB instance running on the localhost interface (e.g. 127.0.0.1) and on the default port (27017). If you want to restore to a different host or port, use the --host (page 1048) and --port (page 1048) options. Consider the following example:
mongorestore --host mongodb1.example.net --port 3017 --username user --password pass /opt/backup/mong
As above, you may specify username and password connections if your mongod (page 1021) requires authentication.
4.2.2 Use Filesystem Snapshots to Backup and Restore MongoDB Databases

This document describes a procedure for creating backups of MongoDB systems using system-level tools, such as LVM or storage appliance, as well as the corresponding restoration strategies. These lesystem snapshots, or block-level backup methods use system level tools to create copies of the device that holds MongoDBs data les. These methods complete quickly and work reliably, but require more system conguration outside of MongoDB. See also: Backup Strategies for MongoDB Systems (page 41) and Use mongodump and mongorestore to Backup and Restore MongoDB Databases (page 43). Snapshots Overview Snapshots work by creating pointers between the live data and a special snapshot volume. These pointers are theoretically equivalent to hard links. As the working data diverges from the snapshot, the snapshot process uses a copy-on-write strategy. As a result the snapshot only stores modied data. After making the snapshot, you mount the snapshot image on your le system and copy data from the snapshot. The resulting backup contains a full copy of all data. Snapshots have the following limitations: The database must be in a consistent or recoverable state when the snapshot takes place. This means that all writes accepted by the database need to be fully written to disk: either to the journal or to data les. If all writes are not on disk when the backup occurs, the backup will not reect these changes. If writes are in progress when the backup occurs, the data les will reect an inconsistent state. With journaling all data-le states resulting from in-progress writes are recoverable; without journaling you must ush all pending writes to disk before running the backup operation and must ensure that no writes occur during the entire backup procedure. If you do use journaling, the journal must reside on the same volume as the data. Snapshots create an image of an entire disk image. Unless you need to back up your entire system, consider isolating your MongoDB data les, journal (if applicable), and conguration on one logical disk that doesnt contain any other data.
46
Alternately, store all MongoDB data les on a dedicated device so that you can make backups without duplicating extraneous data. Ensure that you copy data from snapshots and onto other systems to ensure that data is safe from site failures. Although different snapshots methods provide different capability, the LVM method outlined below does not provide any capacity for capturing incremental backups.
Snapshots With Journaling
If your mongod (page 1021) instance has journaling enabled, then you can use any kind of le system or volume/block level snapshot tool to create backups. If you manage your own infrastructure on a Linux-based system, congure your system with LVM to provide your disk packages and provide snapshot capability. You can also use LVM-based setups within a cloud/virtualized environment. Note: Running LVM provides additional exibility and enables the possibility of using snapshots to back up MongoDB.
Snapshots with Amazon EBS in a RAID 10 Conguration
If your deployment depends on Amazons Elastic Block Storage (EBS) with RAID congured within your instance, it is impossible to get a consistent state across all disks using the platforms snapshot tool. As an alternative, you can do one of the following: Flush all writes to disk and create a write lock to ensure consistent state during the backup process. If you choose this option see Create Backups on Instances that do not have Journaling Enabled (page 49). Congure LVM to run and hold your MongoDB data les on top of the RAID within your system. If you choose this option, perform the LVM backup operation described in Create a Snapshot (page 47). Backup and Restore Using LVM on a Linux System This section provides an overview of a simple backup process using LVM on a Linux system. While the tools, commands, and paths may be (slightly) different on your system the following steps provide a high level overview of the backup operation. Note: Only use the following procedure as a guideline for a backup system and infrastructure. Production backup systems must consider a number of application specic requirements and factors unique to specic environments.
Create a Snapshot
To create a snapshot with LVM , issue a command as root in the following format:
lvcreate --size 100M --snapshot --name mdb-snap01 /dev/vg0/mongodb
This command creates an LVM snapshot (with the --snapshot option) named mdb-snap01 of the mongodb volume in the vg0 volume group.
This example creates a snapshot named mdb-snap01 located at http://docs.mongodb.org/manual/dev/vg0/mdb-snap0 The location and paths to your systems volume groups and devices may vary slightly depending on your operating systems LVM conguration. 4.2. Backup and Recovery Procedures 47
The snapshot has a cap of at 100 megabytes, because of the parameter --size 100M. This size does not reect the total amount of the data on the disk, but rather the quantity of differences between the current state of http://docs.mongodb.org/manual/dev/vg0/mongodb and the creation of the snapshot (i.e. http://docs.mongodb.org/manual/dev/vg0/mdb-snap01.) Warning: Ensure that you create snapshots with enough space to account for data growth, particularly for the period of time that it takes to copy data out of the system or to a temporary image. If your snapshot runs out of space, the snapshot image becomes unusable. Discard this logical volume and create another. The snapshot will exist when the command returns. You can restore directly from the snapshot at any time or by creating a new logical volume and restoring from this snapshot to the alternate image. While snapshots are great for creating high quality backups very quickly, they are not ideal as a format for storing backup data. Snapshots typically depend and reside on the same storage infrastructure as the original disk images. Therefore, its crucial that you archive these snapshots and store them elsewhere.
Archive a Snapshot
After creating a snapshot, mount the snapshot and move the data to separate storage. Your system might try to compress the backup images as you move the ofine. The following procedure fully archives the data from the snapshot:
umount /dev/vg0/mdb-snap01 dd if=/dev/vg0/mdb-snap01 | gzip > mdb-snap01.gz
The above command sequence does the following: Ensures that the http://docs.mongodb.org/manual/dev/vg0/mdb-snap01 device is not mounted. Performs a block level copy of the entire snapshot image using the dd command and compresses the result in a gzipped le in the current working directory. Warning: This command will create a large gz le in your current working directory. Make sure that you run this command in a le system that has enough free space.
Restore a Snapshot
To restore a snapshot created with the above method, issue the following sequence of commands:
lvcreate --size 1G --name mdb-new vg0 gzip -d -c mdb-snap01.gz | dd of=/dev/vg0/mdb-new mount /dev/vg0/mdb-new /srv/mongodb
The above sequence does the following: Creates a new logical volume named mdb-new, in the http://docs.mongodb.org/manual/dev/vg0 volume group. The path to the new device will be http://docs.mongodb.org/manual/dev/vg0/mdb-new. Warning: This volume will have a maximum size of 1 gigabyte. The original le system must have had a total size of 1 gigabyte or smaller, or else the restoration will fail. Change 1G to your desired volume size. Uncompresses and unarchives the mdb-snap01.gz into the mdb-new disk image.
48
Mounts the mdb-new disk image to the http://docs.mongodb.org/manual/srv/mongodb directory. Modify the mount point to correspond to your MongoDB data le location, or other location as needed. Note: The restored snapshot will have a stale mongod.lock le. If you do not remove this le from the snapshot, and MongoDB may assume that the stale lock le indicates an unclean shutdown. If youre running with journal (page 1083) enabled, and you do not use db.fsyncLock() (page 977), you do not need to remove the mongod.lock le. If you use db.fsyncLock() (page 977) you will need to remove the lock.
Restore Directly from a Snapshot
To restore a backup without writing to a compressed gz le, use the following sequence of commands:
umount /dev/vg0/mdb-snap01 lvcreate --size 1G --name mdb-new vg0 dd if=/dev/vg0/mdb-snap01 of=/dev/vg0/mdb-new mount /dev/vg0/mdb-new /srv/mongodb
Remote Backup Storage
You can implement off-system backups using the combined process (page 49) and SSH. This sequence is identical to procedures explained above, except that it archives and compresses the backup on a remote system using SSH. Consider the following procedure:
umount /dev/vg0/mdb-snap01 dd if=/dev/vg0/mdb-snap01 | ssh username@example.com gzip > /opt/backup/mdb-snap01.gz lvcreate --size 1G --name mdb-new vg0 ssh username@example.com gzip -d -c /opt/backup/mdb-snap01.gz | dd of=/dev/vg0/mdb-new mount /dev/vg0/mdb-new /srv/mongodb
Create Backups on Instances that do not have Journaling Enabled If your mongod (page 1021) instance does not run with journaling enabled, or if your journal is on a separate volume, obtaining a functional backup of a consistent state is more complicated. As described in this section, you must ush all writes to disk and lock the database to prevent writes during the backup process. If you have a replica set conguration, then for your backup use a secondary which is not receiving reads (i.e. hidden member). 1. To ush writes to disk and to lock the database (to prevent further writes), issue the db.fsyncLock() (page 977) method in the mongo (page 1036) shell:
db.fsyncLock();
2. Perform the backup operation described in Create a Snapshot (page 47). 3. To unlock the database after the snapshot has completed, use the following command in the mongo (page 1036) shell:
db.fsyncUnlock();
Note: Changed in version 2.0: MongoDB 2.0 added db.fsyncLock() (page 977) and db.fsyncUnlock() (page 977) helpers to the mongo (page 1036) shell. Prior to this version, use the fsync (page 865) command with the lock option, as follows:
49
db.runCommand( { fsync: 1, lock: true } ); db.runCommand( { fsync: 1, lock: false } );
Note: The database cannot be locked with db.fsyncLock() (page 977) while proling is enabled. You must disable proling before locking the database with db.fsyncLock() (page 977). Disable proling using db.setProfilingLevel() (page 985) as follows in the mongo (page 1036) shell:
db.setProfilingLevel(0)
Warning: Changed in version 2.2: When used in combination with fsync (page 865) or db.fsyncLock() (page 977), mongod (page 1021) may block some reads, including those from mongodump (page 1044), when queued write operation waits behind the fsync (page 865) lock.
4.2.3 Copy Databases Between Instances

Synopsis MongoDB provides the copydb (page 862) and clone (page 857) database commands to support migrations of entire logical databases between mongod (page 1021) instances. With these commands you can copy data between instances with a simple interface without the need for an intermediate stage. The db.cloneDatabase() (page 969) and db.copyDatabase() (page 970) provide helpers for these operations in the mongo (page 1036) shell. Data migrations that require an intermediate stage or that involve more than one database instance are beyond the scope of this tutorial. copydb (page 862) and clone (page 857) are more ideal for use cases that resemble the following use cases: data migrations, data warehousing, and seeding test environments. Also consider the Backup Strategies for MongoDB Systems (page 41) and Import and Export MongoDB Data (page 105) documentation for more related information. Note: copydb (page 862) and clone (page 857) do not produce point-in-time snapshots of the source database. Write trafc to the source or destination database during the copy process will result divergent data sets.
Considerations You must run copydb (page 862) or clone (page 857) on the destination server. You cannot use copydb (page 862) or clone (page 857) with databases that have a sharded collection in a sharded cluster, or any database via a mongos (page 1032). You can use copydb (page 862) or clone (page 857) with databases that do not have sharded collections in a cluster when youre connected directly to the mongod (page 1021) instance. You can run copydb (page 862) or clone (page 857) commands on a secondary member of a replica set, with properly congured read preference. Each destination mongod (page 1021) instance must have enough free disk space on the destination server for the database you are copying. Use the db.stats() (page 986) operation to check the size of the database on the source mongod (page 1021) instance. For more information, see db.stats() (page 986). 50 Chapter 4. Backup and Recovery Operations for MongoDB
Processes
Copy and Rename a Database
To copy a database from one MongoDB instance to another and rename the database in the process, use the copydb (page 862) command, or the db.copyDatabase() (page 970) helper in the mongo (page 1036) shell. Use the following procedure to copy the database named test on server db0.example.net to the server named db1.example.net and rename it to records in the process: Verify that the database, test exists on the source mongod (page 1021) instance running on the db0.example.net host. Connect to the destination server, running on the db1.example.net host, using the mongo (page 1036) shell. Model your operation on the following command:
db.copyDatabase( "test", "records", db0.example.net )
Rename a Database
You can also use copydb (page 862) or the db.copyDatabase() (page 970) helper to: rename a database within a single MongoDB instance or create a duplicate database for testing purposes. Use the following procedure to rename the test database records on a single mongod (page 1021) instance: Connect to the mongod (page 1021) using the mongo (page 1036) shell. Model your operation on the following command:
db.copyDatabase( "test", "records" )
Copy a Database with Authentication
To copy a database from a source MongoDB instance that has authentication enabled, you can specify authentication credentials to the copydb (page 862) command or the db.copyDatabase() (page 970) helper in the mongo (page 1036) shell. In the following operation, you will copy the test database from the mongod (page 1021) running on db0.example.net to the records database on the local instance (e.g. db1.example.net.) Because the mongod (page 1021) instance running on db0.example.net requires authentication for all connections, you will need to pass db.copyDatabase() (page 970) authentication credentials, as in the following procedure: Connect to the destination mongod (page 1021) instance running on the db1.example.net host using the mongo (page 1036) shell. Issue the following command:
db.copyDatabase( "test", "records", db0.example.net, "<username>", "<password>")
Replace <username> and <password> with your authentication credentials.
51
Clone a Database
The clone (page 857) command copies a database between mongod (page 1021) instances like copydb (page 862); however, clone (page 857) preserves the database name from the source instance on the destination mongod (page 1021). For many operations, clone (page 857) is functionally equivalent to copydb (page 862), but it has a more simple syntax and a more narrow use. The mongo (page 1036) shell provides the db.cloneDatabase() (page 969) helper as a wrapper around clone (page 857). You can use the following procedure to clone a database from the mongod (page 1021) instance running on db0.example.net to the mongod (page 1021) running on db1.example.net: Connect to the destination mongod (page 1021) instance running on the db1.example.net host using the mongo (page 1036) shell. Issue the following command to specify the name of the database you want to copy:
use records
Use the following operation to initiate the clone (page 857) operation:
db.cloneDatabase( "db0.example.net" )
4.2.4 Recover MongoDB Data following Unexpected Shutdown

If MongoDB does not shutdown cleanly 1 the on-disk representation of the data les will likely reect an inconsistent state which could lead to data corruption. 2 To prevent data inconsistency and corruption, always shut down the database cleanly and use the durability journaling (page 1083). MongoDB writes data to the journal, by default, every 100 milliseconds, such that MongoDB can always recover to a consistent state even in the case of an unclean shutdown due to power loss or other system failure. If you are not running as part of a replica set and do not have journaling enabled, use the following procedure to recover data that may be in an inconsistent state. If you are running as part of a replica set, you should always restore from a backup or restart the mongod (page 1021) instance with an empty dbpath (page 1081) and allow MongoDB to perform an initial sync to restore the data. See also: The Administration (page 33) documents, including Replica Set Syncing (page 411), and the documentation on the repair (page 1085), repairpath (page 1085), and journal (page 1083) settings. Process
Indications
When you are aware of a mongod (page 1021) instance running without journaling that stops unexpectedly and youre not running with replication, you should always run the repair operation before starting MongoDB again. If youre using replication, then restore from a backup and allow replication to perform an initial sync (page 411) to restore data.
1 To ensure a clean shut down, use the mongod --shutdown (page 1028) option, your control script, Control-C (when running mongod (page 1021) in interactive mode,) or kill $(pidof mongod) or kill -2 $(pidof mongod). 2 You can also use the db.collection.validate() (page 950) method to test the integrity of a single collection. However, this process is time consuming, and without journaling you can safely assume that the data is in an invalid state and you should either run the repair operation or resync from an intact member of the replica set.
52
If the mongod.lock le in the data directory specied by dbpath (page 1081), http://docs.mongodb.org/manual/data/db by default, is not a zero-byte le, then mongod (page 1021) will refuse to start, and you will nd a message that contains the following line in your MongoDB log our output:
Unclean shutdown detected.
This indicates that you need to remove the lockle and run repair. If you run repair when the mongodb.lock le exists without the mongod --repairpath (page 1027) option, you will see a message that contains the following line:
old lock file: /data/db/mongod.lock. probably means unclean shutdown
You must remove the lockle and run the repair operation before starting the database normally using the following procedure:
Overview
Warning: Recovering a member of a replica set. Do not use this procedure to recover a member of a replica set. Instead you should either restore from a backup (page 41) or perform an initial sync using data from an intact member of the set, as described in Resync a Member of a Replica Set (page 428). There are two processes to repair data les that result from an unexpected shutdown: 1. Use the --repair (page 1026) option in conjunction with the --repairpath (page 1027) option. mongod (page 1021) will read the existing data les, and write the existing data to new data les. This does not modify or alter the existing data les. You do not need to remove the mongod.lock le before using this procedure. 2. Use the --repair (page 1026) option. mongod (page 1021) will read the existing data les, write the existing data to new les and replace the existing, possibly corrupt, les with new les. You must remove the mongod.lock le before using this procedure. Note: --repair (page 1026) functionality is also available in the shell with the db.repairDatabase() (page 984) helper for the repairDatabase (page 868) command.
Procedures
To repair your data les using the --repairpath (page 1027) option to preserve the original data les unmodied: 1. Start mongod (page 1021) using --repair (page 1026) to read the existing data les.
mongod --dbpath /data/db --repair --repairpath /data/db0
When this completes, the new repaired data http://docs.mongodb.org/manual/data/db0 directory.
les
will
be
in
the
2. Start mongod (page 1021) using the following invocation to point the dbpath (page 1081) at http://docs.mongodb.org/manual/data/db0:
mongod --dbpath /data/db0
Once you conrm that the data les are operational you may delete or archive the data les in the http://docs.mongodb.org/manual/data/db directory.
53
To repair your data les without preserving the original les, do not use the --repairpath (page 1027) option, as in the following procedure: 1. Remove the stale lock le:
rm /data/db/mongod.lock
Replace http://docs.mongodb.org/manual/data/db with your dbpath (page 1081) where your MongoDB instances data les reside. Warning: After you remove the mongod.lock le you must run the --repair (page 1026) process before using your database. 2. Start mongod (page 1021) using --repair (page 1026) to read the existing data les.
mongod --dbpath /data/db --repair
When this completes, the repaired data les will replace http://docs.mongodb.org/manual/data/db directory.
the
original
data
les
in
the
3. Start mongod (page 1021) using the following invocation to point the dbpath (page 1081) at http://docs.mongodb.org/manual/data/db:
mongod --dbpath /data/db
mongod.lock In normal operation, you should never remove the mongod.lock le and start mongod (page 1021). Instead consider the one of the above methods to recover the database and remove the lock les. In dire situations you can remove the lockle, and start the database using the possibly corrupt les, and attempt to recover data from the database; however, its impossible to predict the state of the database in these situations. If you are not running with journaling, and your database shuts down unexpectedly for any reason, you should always proceed as if your database is in an inconsistent and likely corrupt state. If at all possible restore from backup (page 41) or, if running as a replica set, restore by performing an initial sync using data from an intact member of the set, as described in Resync a Member of a Replica Set (page 428).
4.3 Backup and Restore Sharded Clusters

4.3.1 Backup a Small Sharded Cluster with mongodump
Overview If your sharded cluster holds a small data set, you can connect to a mongos (page 1032) using mongodump (page 1044). You can create backups of your MongoDB cluster, if your backup infrastructure can capture the entire backup in a reasonable amount of time and if you have a storage system that can hold the complete MongoDB data set. Read Sharded Cluster Backup Considerations (page 42) for a high-level overview of important considerations as well as a list of alternate backup tutorials. Important: By default mongodump (page 1044) issue its queries to the non-primary nodes.
54
Procedure
Capture Data
Note: If you use mongodump (page 1044) without specifying a database or collection, mongodump (page 1044) will capture collection data and the cluster meta-data from the cong servers (page 500). You cannot use the --oplog (page 1046) option for mongodump (page 1044) when capturing data from mongos (page 1032). This option is only available when running directly against a replica set member. You can perform a backup of a sharded cluster by connecting mongodump (page 1044) to a mongos (page 1032). Use the following operation at your systems prompt:
mongodump --host mongos3.example.net --port 27017
mongodump (page 1044) will write BSON les that hold a copy of data stored in the sharded cluster accessible via the mongos (page 1032) listening on port 27017 of the mongos3.example.net host.
Restore Data
Backups created with mongodump (page 1044) do not reect the chunks or the distribution of data in the sharded collection or collections. Like all mongodump (page 1044) output, these backups contain separate directories for each database and BSON les for each collection in that database. You can restore mongodump (page 1044) output to any MongoDB instance, including a standalone, a replica set, or a new sharded cluster. When restoring data to sharded cluster, you must deploy and congure sharding before restoring data from the backup. See Deploy a Sharded Cluster (page 503) for more information.
4.3.2 Create Backup of a Sharded Cluster with Filesystem Snapshots

Overview This document describes a procedure for taking a backup of all components of a sharded cluster. This procedure uses le system snapshots to capture a copy of the mongod (page 1021) instance. An alternate procedure that uses mongodump (page 1044) to create binary database dumps when le-system snapshots are not available. See Create Backup of a Sharded Cluster with Database Dumps (page 56) for the alternate procedure. See Sharded Cluster Backup Considerations (page 42) for a full higher level overview backing up a sharded cluster as well as links to other tutorials that provide alternate procedures. Important: To capture a point-in-time backup from a sharded cluster you must stop all writes to the cluster. On a running production system, you can only capture an approximation of point-in-time snapshot.
Procedure In this procedure, you will stop the cluster balancer and take a backup up of the cong database, and then take backups of each shard in the cluster using a le-system snapshot tool. If you need an exact moment-in-time snapshot of the system, you will need to stop all application writes before taking the lesystem snapshots; otherwise the snapshot will only approximate a moment in time. For approximate point-in-time snapshots, you can improve the quality of the backup while minimizing impact on the cluster by taking the backup from a secondary member of the replica set that provides each shard.
4.3. Backup and Restore Sharded Clusters
55
1. Disable the balancer process that equalizes the distribution of data among the shards. To disable the balancer, use the sh.stopBalancer() (page 1004) method in the mongo (page 1036) shell, and see the Disable the Balancer (page 529) procedure. Warning: It is essential that you stop the balancer before creating backups. If the balancer remains active, your resulting backups could have duplicate data or miss some data, as chunks may migrate while recording backups. 2. Lock one member of each replica set in each shard so that your backups reect the state of your database at the nearest possible approximation of a single moment in time. Lock these mongod (page 1021) instances in as short of an interval as possible. To lock or freeze a sharded cluster, you must: use the db.fsyncLock() (page 977) method in the mongo (page 1036) shell connected to a single secondary member of the replica set that provides shard mongod (page 1021) instance. Shutdown one of the cong servers (page 500), to prevent all metadata changes during the backup process. 3. Use mongodump (page 1044) to backup one of the cong servers (page 500). This backs up the clusters metadata. You only need to back up one cong server, as they all hold the same data. Issue this command against one of the cong mongod (page 1021) instances or via the mongos (page 1032):
mongodump --db config
4. Back up the replica set members of the shards that you locked. You may back up the shards in parallel. For each shard, create a snapshot. Use the procedures in Use Filesystem Snapshots to Backup and Restore MongoDB Databases (page 46). 5. Unlock all locked replica set members of each shard using the db.fsyncUnlock() (page 977) method in the mongo (page 1036) shell. 6. Restore the balancer with the sh.startBalancer() (page 1002) method according to the Disable the Balancer (page 529) procedure. Use the following command sequence when connected to the mongos (page 1032) with the mongo (page 1036) shell:
use config sh.startBalancer()
4.3.3 Create Backup of a Sharded Cluster with Database Dumps

Overview This document describes a procedure for taking a backup of all components of a sharded cluster. This procedure uses mongodump (page 1044) to create dumps of the mongod (page 1021) instance. An alternate procedure uses le system snapshots to capture the backup data, and may be more efcient in some situations if your system conguration allows le system backups. See Create Backup of a Sharded Cluster with Filesystem Snapshots (page 55). See Sharded Cluster Backup Considerations (page 42) for a full higher level overview of backing up a sharded cluster as well as links to other tutorials that provide alternate procedures. Important: To capture a point-in-time backup from a sharded cluster you must stop all writes to the cluster. On a running production system, you can only capture an approximation of point-in-time snapshot.
56
Procedure In this procedure, you will stop the cluster balancer and take a backup up of the cong database, and then take backups of each shard in the cluster using mongodump (page 1044) to capture the backup data. If you need an exact momentin-time snapshot of the system, you will need to stop all application writes before taking the lesystem snapshots; otherwise the snapshot will only approximate a moment of time. For approximate point-in-time snapshots, you can improve the quality of the backup while minimizing impact on the cluster by taking the backup from a secondary member of the replica set that provides each shard. 1. Disable the balancer process that equalizes the distribution of data among the shards. To disable the balancer, use the sh.stopBalancer() (page 1004) method in the mongo (page 1036) shell, and see the Disable the Balancer (page 529) procedure. Warning: It is essential that you stop the balancer before creating backups. If the balancer remains active, your resulting backups could have duplicate data or miss some data, as chunks migrate while recording backups. 2. Lock one member of each replica set in each shard so that your backups reect the state of your database at the nearest possible approximation of a single moment in time. Lock these mongod (page 1021) instances in as short of an interval as possible. To lock or freeze a sharded cluster, you must: Shutdown one member of each replica set. Ensure that the oplog has sufcient capacity to allow these secondaries to catch up to the state of the primaries after nishing the backup procedure. See Oplog (page 392) for more information. Shutdown one of the cong servers (page 500), to prevent all metadata changes during the backup process. 3. Use mongodump (page 1044) to backup one of the cong servers (page 500). This backs up the clusters metadata. You only need to back up one cong server, as they all hold the same data. Issue this command against one of the cong mongod (page 1021) instances or via the mongos (page 1032):
mongodump --journal --db config
4. Back up the replica set members of the shards that shut down using mongodump (page 1044) and specifying the --dbpath (page 1045) option. You may back up the shards in parallel. Consider the following invocation:
mongodump --journal --dbpath /data/db/ --out /data/backup/
You must run this command on the system where the mongod (page 1021) ran. This operation will use journaling and create a dump of the entire mongod (page 1021) instance with data les stored in http://docs.mongodb.org/manual/data/db/. mongodump (page 1044) will write the output of this dump to the http://docs.mongodb.org/manual/data/backup/ directory. 5. Restart all stopped replica set members of each shard as normal and allow them to catch up with the state of the primary. 6. Restore the balancer with the sh.startBalancer() (page 1002) method according to the Disable the Balancer (page 529) procedure. Use the following command sequence when connected to the mongos (page 1032) with the mongo (page 1036) shell:
use config sh.startBalancer()
57
4.3.4 Restore a Single Shard

Overview Restoring a single shard from backup with other unaffected shards requires a number of special considerations and practices. This document outlines the additional tasks you must perform when restoring a single shard. Consider the following resources on backups in general as well as backup and restoration of sharded clusters specically: Sharded Cluster Backup Considerations (page 42) Restore Sharded Clusters (page 58) Backup Strategies for MongoDB Systems (page 41) Procedure Always restore sharded clusters as a whole. When you restore a single shard, keep in mind that the balancer process might have moved chunks to or from this shard since the last backup. If thats the case, you must manually move those chunks, as described in this procedure. 1. Restore the shard as you would any other mongod (page 1021) instance. See Backup Strategies for MongoDB Systems (page 41) for overviews of these procedures. 2. For all chunks that migrate away from this shard, you do not need to do anything at this time. You do not need to delete these documents from the shard because the chunks are automatically ltered out from queries by mongos (page 1032). You can remove these documents from the shard, if you like, at your leisure. 3. For chunks that migrate to this shard after the most recent backup, you must manually recover the chunks using backups of other shards, or some other source. To determine what chunks have moved, view the changelog collection in the Cong Database (page 545).
4.3.5 Restore Sharded Clusters

Overview The procedure outlined in this document addresses how to restore an entire sharded cluster. For information on related backup procedures consider the following tutorials which describe backup procedures in greater detail: Create Backup of a Sharded Cluster with Filesystem Snapshots (page 55) Create Backup of a Sharded Cluster with Database Dumps (page 56) The exact procedure used to restore a database depends on the method used to capture the backup. See the Backup Strategies for MongoDB Systems (page 41) document for an overview of backups with MongoDB, as well as Sharded Cluster Backup Considerations (page 42) which provides an overview of the high level concepts important for backing up sharded clusters. Procedure 1. Stop all mongod (page 1021) and mongos (page 1032) processes. 2. If shard hostnames have changed, you must manually update the shards collection in the Cong Database (page 545) to use the new hostnames. Do the following:
58
(a) Start the three cong servers (page 500) by issuing commands similar to the following, using values appropriate to your conguration:
mongod --configsvr --dbpath /data/configdb --port 27019
(b) Restore the Cong Database (page 545) on each cong server. (c) Start one mongos (page 1032) instance. (d) Update the Cong Database (page 545) collection named shards to reect the new hostnames. 3. Restore the following: Data les for each server in each shard. Because replica sets provide each production shard, restore all the members of the replica set or use the other standard approaches for restoring a replica set from backup. See the Restore a Snapshot (page 48) and Restore a Database with mongorestore (page 44) sections for details on these procedures. Data les for each cong server (page 500), if you have not already done so in the previous step. 4. Restart all the mongos (page 1032) instances. 5. Restart all the mongod (page 1021) instances. 6. Connect to a mongos (page 1032) instance from a mongo (page 1036) shell and use the db.printShardingStatus() (page 983) method to ensure that the cluster is operational, as follows:
db.printShardingStatus() show collections
4.3.6 Schedule Backup Window for Sharded Clusters

Overview In a sharded cluster, the balancer process is responsible for distributing sharded data around the cluster, so that each shard has roughly the same amount of data. However, when creating backups from a sharded cluster it is important that you disable the balancer while taking backups to ensure that no chunk migrations affect the content of the backup captured by the backup procedure. Using the procedure outlined in the section Disable the Balancer (page 529) you can manually stop the balancer process temporarily. As an alternative you can use this procedure to dene a balancing window so that the balancer is always disabled during your automated backup operation. Procedure If you have an automated backup schedule, you can disable all balancing operations for a period of time. For instance, consider the following command:
use config db.settings.update( { _id : "balancer" }, { $set : { activeWindow : { start : "6:00", stop : "23:00"
This operation congures the balancer to run between 6:00am and 11:00pm, server time. Schedule your backup operation to run and complete outside of this time. Ensure that the backup can complete outside the window when the balancer is running and that the balancer can effectively balance the collection among the shards in the window allotted to each.
59
60
CHAPTER 5
Data Center Awareness
MongoDB provides a number of features that allow application developers and database administrators to customize the behavior of a sharded cluster or replica set deployment so that MongoDB may be more data center aware, or allow operational and location-based separation. MongoDB also supports segregation based on functional parameters, to ensure that certain mongod (page 1021) instances are only used for reporting workloads or that certain high-frequency portions of a sharded collection only exist on specic shards. Consider the following documents:
5.1 Operational Segregation in MongoDB Operations and Deployments

5.1.1 Operational Overview
MongoDB includes a number of features that allow database administrators and developers to segregate application operations to MongoDB deployments by functional or geographical groupings. This capability provides data center awareness, which allows applications to target MongoDB deployments with consideration of the physical location of the mongod (page 1021) instances. MongoDB supports segmentation of operations across different dimensions, which may include multiple data centers and geographical regions in multidata center deployments, racks, networks, or power circuits in single data center deployments. MongoDB also supports segregation of database operations based on functional or operational parameters, to ensure that certain mongod (page 1021) instances are only used for reporting workloads or that certain high-frequency portions of a sharded collection only exist on specic shards. Specically, with MongoDB, you can: ensure write operations propagate to specic members of a replica set, or to specic members of replica sets. ensure that specic members of a replica set respond to queries. ensure that specic ranges of your shard key balance onto and reside on specic shards. combine the above features in a single distributed deployment, on a per-operation (for read and write operations) and collection (for chunk distribution in sharded clusters distribution) basis. For full documentation of these features, see the following documentation in the MongoDB Manual: Read Preferences (page 402), which controls how drivers help applications target read operations to members of a replica set.
61
Write Concerns (page 398), which controls how MongoDB ensures that write operations propagate to members of a replica set. Replica Set Tags (page 455), which control how applications create and interact with custom groupings of replica set members to create custom application-specic read preferences and write concerns. Tag Aware Sharding (page 532), which allows MongoDB administrators to dene an application-specic balancing policy, to control how documents belonging to specic ranges of a shard key distribute to shards in the sharded cluster. See also: Before adding operational segregation features to your application and MongoDB deployment, become familiar with all documentation of replication (page 385) and sharding (page 483), particularly Replica Set Fundamental Concepts (page 387) and Sharded Cluster Overview (page 485).
5.2 Tag Aware Sharding

For sharded clusters, MongoDB makes it possible to associate specic ranges of a shard key with a specic shard or subset of shards. This association dictates the policy of the cluster balancer process as it balances the chunks around the cluster. This capability enables the following deployment patterns: isolating a specic subset of data on specic set of shards. controlling the balancing policy so that, in a geographically distributed cluster, the most relevant portions of the data set reside on the shards with the greatest proximity to the application servers. This document describes the behavior, operation, and use of tag aware sharding in MongoDB deployments. Note: Shard key range tags are entirely distinct from replica set member tags (page 404). Hash-based sharding does not support tag-aware sharding.
5.2.1 Behavior and Operations

Tags in a sharded cluster are pieces of metadata that dictate the policy and behavior of the cluster balancer. Using tags, you may associate individual shards in a cluster with one or more tags. Then, you can assign this tag string to a range of shard key values for a sharded collection. When migrating a chunk, the balancer will select a destination shard based on the congured tag ranges. The balancer migrates chunks in tagged ranges to shards with those tags, if tagged shards are not balanced.
1
Note: Because a single chunk may span different tagged shard key ranges, the balancer may migrate chunks to tagged shards that contain values that exceed the upper bound of the selected tag range. Example Given a sharded collection with two congured tag ranges, such that: Shard key values between 100 and 200 have tags to direct corresponding chunks to shards tagged NYC. Shard Key values between 200 and 300 have tags to direct corresponding chunks to shards tagged SFO.
1 To migrate chunks in a tagged environment, the balancer selects a target shard with a tag range that has an upper bound that is greater than the migrating chunks lower bound. If a shard with a matching tagged range exists, the balancer will migrate the chunk to that shard.
62
Chapter 5. Data Center Awareness
In this cluster, the balancer will migrate a chunk with shard key values ranging between 150 and 220 to a shard tagged NYC, since 150 is closer to 200 than 300. After conguring tags on the shards and ranges of the shard key, the cluster may take some time to reach the proper distribution of data, depending on the division of chunks (i.e. splits) and the current distribution of data in the cluster. Once congured, the balancer will respect tag ranges during future balancing rounds (page 497). See also: Administer and Manage Shard Tags (page 517)
5.3 Administer and Manage Shard Tags

In a sharded cluster, you can use tags to associate specic ranges of a shard key with a specic shard or subset of shards.
5.3.1 Tag a Shard

Associate tags with a particular shard using the sh.addShardTag() (page 996) method when connected to a mongos (page 1032) instance. A single shard may have multiple tags, and multiple shards may also have the same tag. Example The following example adds the tag NYC to two shards, and the tags SFO and NRT to a third shard:
sh.addShardTag("shard0000", sh.addShardTag("shard0001", sh.addShardTag("shard0002", sh.addShardTag("shard0002", "NYC") "NYC") "SFO") "NRT")
You may remove tags from a particular shard using the sh.removeShardTag() (page 1000) method when connected to a mongos (page 1032) instance, as in the following example, which removes the NRT tag from a shard:
sh.removeShardTag("shard0002", "NRT")
5.3.2 Tag a Shard Key Range

To assign a tag to a range of shard keys use the sh.addTagRange() method when connected to a mongos (page 1032) instance. Any given shard key range may only have one assigned tag. You cannot overlap dened ranges, or tag the same range more than once. Example Given a collection named users in the records database, sharded by the zipcode eld. The following operations assign: two ranges of zip codes in Manhattan and Brooklyn the NYC tag one range of zip codes in San Francisco the SFO tag
sh.addTagRange("records.users", { zipcode: "10001" }, { zipcode: "10281" }, "NYC") sh.addTagRange("records.users", { zipcode: "11201" }, { zipcode: "11240" }, "NYC") sh.addTagRange("records.users", { zipcode: "94102" }, { zipcode: "94135" }, "SFO")
5.3. Administer and Manage Shard Tags
63
Note: Shard ranges are always inclusive of the lower value and exclusive of the upper boundary.
5.3.3 Remove a Tag From a Shard Key Range

The mongod (page 1021) does not provide a helper for removing a tag range. You may delete tag assignment from a shard key range by removing the corresponding document from the tags (page 550) collection of the config database. Each document in the tags (page 550) holds the namespace of the sharded collection and a minimum shard key value. Example The following example removes the NYC tag assignment for the range of zip codes within Manhattan:
use config db.tags.remove({ _id: { ns: "records.users", min: { zipcode: "10001" }}, tag: "NYC" })
5.3.4 View Existing Shard Tags

The output from sh.status() (page 1002) lists tags associated with a shard, if any, for each shard. A shards tags exist in the shards document in the shards (page 549) collection of the config database. To return all shards with a specic tag, use a sequence of operations that resemble the following, which will return only those shards tagged with NYC:
use config db.shards.find({ tags: "NYC" })
You can nd tag ranges for all namespaces in the tags (page 550) collection of the config database. The output of sh.status() (page 1002) displays all tag ranges. To return all shard key ranges tagged with NYC, use the following sequence of operations:
use config db.tags.find({ tags: "NYC" })
5.4 Deploy a Geographically Distributed Replica Set

This tutorial outlines the process for deploying a replica set with members in multiple locations. The tutorial addresses three-member sets, four-member sets, and sets with more than four members. For appropriate background, see Replica Set Fundamental Concepts (page 387) and Replica Set Architectures and Deployment Patterns (page 395). For related tutorials, see Deploy a Replica Set (page 419) and Add Members to a Replica Set (page 424).
5.4.1 Overview
While replica sets provide basic protection against single-instance failure, when all of the members of a replica set reside in a single facility, the replica set is still susceptible to some classes of errors in that facility including power outages, networking distortions, and natural disasters. To protect against these classes of failures, deploy a replica set with one or more members in a geographically distinct facility or data center. 64 Chapter 5. Data Center Awareness
5.4.2 Requirements
For a three-member replica set you need two instances in a primary facility (hereafter, Site A) and one member in a secondary facility (hereafter, Site B.) Site A should be the same facility or very close to your primary application infrastructure (i.e. application servers, caching layer, users, etc.) For a four-member replica set you need two members in Site A, two members in Site B (or one member in Site B and one member in Site C,) and a single arbiter in Site A. For replica sets with additional members in the secondary facility or with multiple secondary facilities, the requirements are the same as above but with the following notes: Ensure that a majority of the voting members (page 389) are within Site A. This includes secondary-only members (page 388) and arbiters (page 388) For more information on the need to keep the voting majority on one site, see Elections (page 389). If you deploy a replica set with an uneven number of members, deploy an arbiter (page 388) on Site A. The arbiter must be on site A to keep the majority there. For all congurations in this tutorial, deploy each replica set member on a separate system. Although you may deploy more than one replica set member on a single system, doing so reduces the redundancy and capacity of the replica set. Such deployments are typically for testing purposes and beyond the scope of this tutorial.
5.4.3 Procedures
Deploy a Distributed Three-Member Replica Set A geographically distributed three-member deployment has the following features: Each member of the replica set resides on its own machine, and the MongoDB processes all bind to port 27017, which is the standard MongoDB port. Each member of the replica set must be accessible by way of resolvable DNS or hostnames in the following scheme: mongodb0.example.net mongodb1.example.net mongodb2.example.net Congure DNS names appropriately, or set up your systems http://docs.mongodb.org/manual/etc/hosts le to reect this conguration. Ensure that one system (e.g. mongodb2.example.net) resides in Site B. Host all other systems in Site A. Ensure that network trafc can pass between all members in the network securely and efciently. Consider the following: Establish a virtual private network between the systems in Site A and Site B to encrypt all trafc between the sites and remains private. Ensure that your network topology routes all trafc between members within a single site over the local area network. Congure authentication using auth (page 1081) and keyFile (page 1081), so that only servers and process with authentication can connect to the replica set. Congure networking and rewall rules so that only trafc (incoming and outgoing packets) on the default MongoDB port (e.g. 27017) from within your deployment. See also: For more information on security and rewalls, see Security (page 393).
5.4. Deploy a Geographically Distributed Replica Set
65
Specify run-time conguration on each system in a conguration le (page 1078) stored in http://docs.mongodb.org/manual/etc/mongodb.conf or in a related location. Do not specify run-time conguration through command line options. For each MongoDB instance, use the following conguration, with values set appropriate to your systems:
port = 27017 bind_ip = 10.8.0.10 dbpath = /srv/mongodb/ fork = true replSet = rs0/mongodb0.example.net,mongodb1.example.net,mongodb2.example.net
Modify bind_ip (page 1079) to reect a secure interface on your system that is able to access all other members of the set and that is accessible to all other members of the replica set. The DNS or host names need to point and resolve to this IP address. Congure network rules or a virtual private network (i.e. VPN) to permit this access. Note: The portion of the replSet (page 1088) following the http://docs.mongodb.org/manual/ provides a seed list of known members of the replica set. mongod (page 1021) uses this list to fetch conguration changes following restarts. It is acceptable to omit this section entirely, and have the replSet (page 1088) option resemble:
replSet = rs0
For more documentation on the above run time congurations, as well as additional conguration options, see Conguration File Options (page 1078). To deploy a geographically distributed three-member set: 1. On each system start the mongod (page 1021) process by issuing a command similar to following:
mongod --config /etc/mongodb.conf
Note: In production deployments you likely want to use and congure a control script to manage this process based on this command. Control scripts are beyond the scope of this document. 2. Open a mongo (page 1036) shell connected to one of the mongod (page 1021) instances:
mongo
3. Use the rs.initiate() (page 989) method on one member to initiate a replica set consisting of the current member and using the default conguration:
rs.initiate()
4. Display the current replica conguration (page 463):

rs.conf()
5. Add the remaining members to the replica set by issuing a sequence of commands similar to the following. The example commands assume the current primary is mongodb0.example.net:
rs.add("mongodb1.example.net") rs.add("mongodb2.example.net")
66
6. Make sure that you have congured the member located in Site B (i.e. mongodb2.example.net) as a secondary-only member (page 388): (a) Issue the following command to determine the _id (page 463) value for mongodb2.example.net:
rs.conf()
(b) In the members (page 463) array, save the _id (page 463) value. The example in the next step assumes this value is 2. (c) In the mongo (page 1036) shell connected to the replica sets primary, issue a command sequence similar to the following:
cfg = rs.conf() cfg.members[2].priority = 0 rs.reconfig(cfg)
Note: In some situations, the rs.reconfig() (page 989) shell method can force the current primary to step down and causes an election. When the primary steps down, all clients will disconnect. This is the intended behavior. While, this typically takes 10-20 seconds, attempt to make these changes during scheduled maintenance periods. After these commands return you have a geographically distributed three-member replica set. 7. To check the status of your replica set, issue rs.status() (page 990). See also: The documentation of the following shell functions for more information: rs.initiate() (page 989) rs.conf() (page 988) rs.reconfig() (page 989) rs.add() (page 987) Deploy a Distributed Four-Member Replica Set A geographically distributed four-member deployment has the following features: Each member of the replica set, except for the arbiter (see below), resides on its own machine, and the MongoDB processes all bind to port 27017, which is the standard MongoDB port. Each member of the replica set must be accessible by way of resolvable DNS or hostnames in the following scheme: mongodb0.example.net mongodb1.example.net mongodb2.example.net mongodb3.example.net Congure DNS names appropriately, or set up your systems http://docs.mongodb.org/manual/etc/host le to reect this conguration. Ensure that one system (e.g. mongodb2.example.net) resides in Site B. Host all other systems in Site A. One host (e.g. mongodb3.example.net) will be an arbiter and can run on a system that is also used for an application server or some other shared purpose.
67
There are three possible architectures for this replica set: Two members in Site A, two secondary-only members (page 388) in Site B, and an arbiter in Site A. Three members in Site A and one secondary-only member in Site B. Two members in Site A, one secondary-only member in Site B, one secondary-only member in Site C, and an arbiter in site A. In most cases the rst architecture is preferable because it is the least complex. Ensure that network trafc can pass between all members in the network securely and efciently. Consider the following: Establish a virtual private network between the systems in Site A and Site B (and Site C if it exists) to encrypt all trafc between the sites and remains private. Ensure that your network topology routes all trafc between members within a single site over the local area network. Congure authentication using auth (page 1081) and keyFile (page 1081), so that only servers and process with authentication can connect to the replica set. Congure networking and rewall rules so that only trafc (incoming and outgoing packets) on the default MongoDB port (e.g. 27017) from within your deployment. See also: For more information on security and rewalls, see Security (page 393). Specify run-time conguration on each system in a conguration le (page 1078) stored in http://docs.mongodb.org/manual/etc/mongodb.conf or in a related location. Do not specify run-time conguration through command line options. For each MongoDB instance, use the following conguration, with values set appropriate to your systems:
port = 27017 bind_ip = 10.8.0.10 dbpath = /srv/mongodb/ fork = true
replSet = rs0/mongodb0.example.net,mongodb1.example.net,mongodb2.example.net,mongodb3.example.ne
replSet = rs0
For more documentation on the above run time congurations, as well as additional conguration options, see doc:/reference/conguration-options. To deploy a geographically distributed four-member set: 1. On each system start the mongod (page 1021) process by issuing a command similar to following:
68
Note: In production deployments you likely want to use and congure a control script to manage this process based on this command. Control scripts are beyond the scope of this document. 2. Open a mongo (page 1036) shell connected to this host:
mongo
3. Use rs.initiate() (page 989) to initiate a replica set consisting of the current member and using the default conguration:
rs.initiate()

rs.conf()
rs.add("mongodb1.example.net") rs.add("mongodb2.example.net") rs.add("mongodb3.example.net")
6. In the same shell session, issue the following command to add the arbiter (e.g. mongodb4.example.net):
rs.addArb("mongodb4.example.net")
7. Make sure that you have congured each member located in Site B (e.g. mongodb3.example.net) as a secondary-only member (page 388): (a) Issue the following command to determine the _id (page 463) value for the member:
rs.conf()
Note: In some situations, the rs.reconfig() (page 989) shell method can force the current primary to step down and causes an election. When the primary steps down, all clients will disconnect. This is the intended behavior. While, this typically takes 10-20 seconds, attempt to make these changes during scheduled maintenance periods. After these commands return you have a geographically distributed four-member replica set. 8. To check the status of your replica set, issue rs.status() (page 990). See also: The documentation of the following shell functions for more information: rs.initiate() (page 989)
69
rs.conf() (page 988) rs.reconfig() (page 989) rs.add() (page 987) Deploy a Distributed Set with More than Four Members The procedure for deploying a geographically distributed set with more than four members is similar to the above procedures, with the following differences: Never deploy more than seven voting members. Use the procedure for a four-member set if you have an even number of members (see Deploy a Distributed Four-Member Replica Set (page 432)). Ensure that Site A always has a majority of the members by deploying the arbiter within Site A. For six member sets, deploy at least three voting members in addition to the arbiter in Site A, the remaining members in alternate sites. Use the procedure for a three-member set if you have an odd number of members (see Deploy a Distributed Three-Member Replica Set (page 430)). Ensure that Site A always has a majority of the members of the set. For example, if a set has ve members, deploy three members within the primary facility and two members in other facilities. If you have a majority of the members of the set outside of Site A and the network partitions to prevent communication between sites, the current primary in Site A will step down, even if none of the members outside of Site A are eligible to become primary. Additionally, consider the Write Concern (page 398) and Read Preference (page 402) documents, which addresses capabilities related to data center awareness.
70
CHAPTER 6
Journaling
MongoDB uses write ahead logging to an on-disk journal to guarantee write operation (page 179) durability and to provide crash resiliency. Before applying a change to the data les, MongoDB writes the change operation to the journal. If MongoDB should terminate or encounter an error before it can write the changes from the journal to the data les, MongoDB can re-apply the write operation and maintain a consistent state. Without a journal, if mongod (page 1021) exits unexpectedly, you must assume your data is in an inconsistent state, and you must run either repair (page 52) or, preferably, resync (page 428) from a clean member of the replica set. With journaling enabled, if mongod (page 1021) stops unexpectedly, the program can recover everything written to the journal, and the data remains in a consistent state. By default, the greatest extent of lost writes, i.e., those not made to the journal, are those made in the last 100 milliseconds. See journalCommitInterval (page 1083) for more information on the default. With journaling, if you want a data set to reside entirely in RAM, you need enough RAM to hold the dataset plus the write working set. The write working set is the amount of unique data you expect to see written between re-mappings of the private view. For information on views, see Storage Views used in Journaling (page 74). Important: Changed in version 2.0: For 64-bit builds of mongod (page 1021), journaling is enabled by default. For other platforms, see journal (page 1083).
6.1 Procedures
6.1.1 Enable Journaling
Changed in version 2.0: For 64-bit builds of mongod (page 1021), journaling is enabled by default. To enable journaling, start mongod (page 1021) with the --journal (page 1025) command line option. If no journal les exist, when mongod (page 1021) starts, it must preallocate new journal les. During this operation, the mongod (page 1021) is not listening for connections until preallocation completes: for some systems this may take a several minutes. During this period your applications and the mongo (page 1036) shell are not available.
6.1.2 Disable Journaling

Warning: Do not disable journaling on production systems. If your mongod (page 1021) instance stops without shutting down cleanly unexpectedly for any reason, (e.g. power failure) and you are not running with journaling, then you must recover from an unaffected replica set member or backup, as described in repair (page 52).
71
To disable journaling, start mongod (page 1021) with the --nojournal (page 1026) command line option.
6.1.3 Get Commit Acknowledgment

You can get commit acknowledgment with the getLastError (page 831) command and the j option. For details, see Write Concern Reference (page 183).
6.1.4 Avoid Preallocation Lag

To avoid preallocation lag (page 73), you can preallocate les in the journal directory by copying them from another instance of mongod (page 1021). Preallocated les do not contain data. It is safe to later remove them. But if you restart mongod (page 1021) with journaling, mongod (page 1021) will create them again. Example The following sequence preallocates journal les for an instance of mongod (page 1021) running on port 27017 with a database path of http://docs.mongodb.org/manual/data/db. For demonstration purposes, the sequence starts by creating a set of journal les in the usual way. 1. Create a temporary directory into which to create a set of journal les:
mkdir ~/tmpDbpath
2. Create a set of journal les by staring a mongod (page 1021) instance that uses the temporary directory:
mongod --port 10000 --dbpath ~/tmpDbpath --journal
3. When you see the following log output, indicating mongod (page 1021) has the les, press CONTROL+C to stop the mongod (page 1021) instance:
web admin interface listening on port 11000
4. Preallocate journal les for the new instance of mongod (page 1021) by moving the journal les from the data directory of the existing instance to the data directory of the new instance:
mv ~/tmpDbpath/journal /data/db/
5. Start the new mongod (page 1021) instance:

mongod --port 27017 --dbpath /data/db --journal
6.1.5 Monitor Journal Status

Use the following commands and methods to monitor journal status: serverStatus (page 889) The serverStatus (page 889) command returns database status information that is useful for assessing performance. journalLatencyTest (page 914) Use journalLatencyTest (page 914) to measure how long it takes on your volume to write to the disk in an append-only fashion. You can run this command on an idle system to get a baseline sync time for journaling.
72
Chapter 6. Journaling
You can also run this command on a busy system to see the sync time on a busy system, which may be higher if the journal directory is on the same volume as the data les. The journalLatencyTest (page 914) command also provides a way to check if your disk drive is buffering writes in its local cache. If the number is very low (i.e., less than 2 milliseconds) and the drive is non-SSD, the drive is probably buffering writes. In that case, enable cache write-through for the device in your operating system, unless you have a disk controller card with battery backed RAM.
6.1.6 Change the Group Commit Interval

Changed in version 2.0. You can set the group commit interval using the --journalCommitInterval (page 1025) command line option. The allowed range is 2 to 300 milliseconds. Lower values increase the durability of the journal at the expense of disk performance.
6.1.7 Recover Data After Unexpected Shutdown

On a restart after a crash, MongoDB replays all journal les in the journal directory before the server becomes available. If MongoDB must replay journal les, mongod (page 1021) notes these events in the log output. There is no reason to run repairDatabase (page 868) in these situations.
6.2 Journaling Internals

When running with journaling, MongoDB stores and applies write operations (page 179) in memory and in the journal before the changes are in the data les.
6.2.1 Journal Files

With journaling enabled, MongoDB creates a journal directory within the directory dened by dbpath (page 1081), which is http://docs.mongodb.org/manual/data/db by default. The journal directory holds journal les, which contain write-ahead redo logs. The directory also holds a last-sequence-number le. A clean shutdown removes all the les in the journal directory. Journal les are append-only les and have le names prexed with j._. When a journal le holds 1 gigabyte of data, MongoDB creates a new journal le. Once MongoDB applies all the write operations in the journal les, it deletes these les. Unless you write many bytes of data per-second, the journal directory should contain only two or three journal les. To limit the size of each journal le to 128 megabytes, use the smallfiles (page 1085) run time option when starting mongod (page 1021). To speed the frequent sequential writes that occur to the current journal le, you can ensure that the journal directory is on a different system. Important: If you place the journal on a different lesystem from your data les you cannot use a lesystem snapshot to capture consistent backups of a dbpath (page 1081) directory. Note: Depending on your le system, you might experience a preallocation lag the rst time you start a mongod (page 1021) instance with journaling enabled. MongoDB preallocates journal les if it is faster on your le system to create les of a pre-dened. The amount of time required to pre-allocate lag might last several minutes, during 6.2. Journaling Internals 73
which you will not be able to connect to the database. This is a one-time preallocation and does not occur with future invocations. To avoid preallocation lag, see Avoid Preallocation Lag (page 72).
6.2.2 Storage Views used in Journaling

Journaling adds three storage views to MongoDB. The shared view stores modied data for upload to the MongoDB data les. The shared view is the only view with direct access to the MongoDB data les. When running with journaling, mongod (page 1021) asks the operating system to map your existing on-disk data les to the shared view memory view. The operating system maps the les but does not load them. MongoDB later loads data les to shared view as needed. The private view stores data for use in read operations (page 167). MongoDB maps private view to the shared view and is the rst place MongoDB applies new write operations (page 179). The journal is an on-disk view that stores new write operations after MongoDB applies the operation to the private cache but before applying them to the data les. The journal provides durability. If the mongod (page 1021) instance were to crash without having applied the writes to the data les, the journal could replay the writes to the shared view for eventual upload to the data les.
6.2.3 How Journaling Records Write Operations

MongoDB copies the write operations to the journal in batches called group commits. See journalCommitInterval (page 1083) for more information on the default commit interval. These group commits help minimize the performance impact of journaling. Journaling stores raw operations that allow MongoDB to reconstruct the following: document insertion/updates index modications changes to the namespace les As write operations (page 179) occur, MongoDB writes the data to the private view in RAM and then copies the write operations in batches to the journal. The journal stores the operations on disk to ensure durability. MongoDB adds the operations as entries on the journals forward pointer. Each entry describes which bytes the write operation changed in the data les. MongoDB next applies the journals write operations to the shared view. At this point, the shared view becomes inconsistent with the data les. At default intervals of 60 seconds, MongoDB asks the operating system to ush the shared view to disk. This brings the data les up-to-date with the latest write operations. When MongoDB ushes write operations to the data les, MongoDB removes the write operations from the journals behind pointer. The behind pointer is always far back from advanced pointer. As part of journaling, MongoDB routinely asks the operating system to remap the shared view to the private view, for consistency. Note: The interaction between the shared view and the on-disk data les is similar to how MongoDB works without journaling, which is that MongoDB asks the operating system to ush in-memory changes back to the data les every 60 seconds.
74
Chapter 6. Journaling
CHAPTER 7
Connect to MongoDB with SSL
This document outlines the use and operation of MongoDBs SSL support. SSL allows MongoDB clients to support encrypted connections to mongod (page 1021) instances. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL, you must either build MongoDB locally passing the --ssl option to scons or use MongoDB Enterprise. These instructions outline the process for getting started with SSL and assume that you have already installed a build of MongoDB that includes SSL support and that your client driver supports SSL.
7.1 Congure mongod and mongos for SSL

7.1.1 Combine SSL Certicate and Key File
Before you can use SSL, you must have a .pem le that contains the public key certicate and private key. MongoDB can use any valid SSL certicate. To generate a self-signed certicate and private key, use a command that resembles the following:
cd /etc/ssl/ openssl req -new -x509 -days 365 -nodes -out mongodb-cert.crt -keyout mongodb-cert.key
This operation generates a new, self-signed certicate with no passphrase that is valid for 365 days. Once you have the certicate, concatenate the certicate and private key to a .pem le, as in the following example:
cat mongodb-cert.key mongodb-cert.crt > mongodb.pem
7.1.2 Set Up mongod and mongos with SSL Certicate and Key
To use SSL in your MongoDB deployment, include the following run-time options with mongod (page 1021) and mongos (page 1032): sslOnNormalPorts (page 1091) sslPEMKeyFile (page 1091) with the .pem le that contains the SSL certicate and key. Consider the following syntax for mongod (page 1021):
mongod --sslOnNormalPorts --sslPEMKeyFile <pem>
75
For example, given an SSL certicate located at http://docs.mongodb.org/manual/etc/ssl/mongodb.pem, congure mongod (page 1021) to use SSL encryption for all connections with the following command:
mongod --sslOnNormalPorts --sslPEMKeyFile /etc/ssl/mongodb.pem
Note: Specify <pem> with the full path name to the certicate. If the private key portion of the <pem> is encrypted, specify the encryption password with the sslPEMKeyPassword (page 1091) option. You may also specify these options in the conguration le (page 1078), as in the following example:
sslOnNormalPorts = true sslPEMKeyFile = /etc/ssl/mongodb.pem
To connect, to mongod (page 1021) and mongos (page 1032) instances using SSL, the mongo (page 1036) shell and MongoDB tools must include the --ssl option. See SSL Conguration for Clients (page 77) for more information on connecting to mongod (page 1021) and mongos (page 1032) running with SSL.
7.1.3 Set Up mongod and mongos with Certicate Validation

To set up mongod (page 1021) or mongos (page 1032) for SSL encryption using an SSL certicate signed by a certicate authority, include the following run-time options during startup: sslOnNormalPorts (page 1091) sslPEMKeyFile (page 1091) with the name of the .pem le that contains the signed SSL certicate and key. sslCAFile (page 1091) with the name of the .pem le that contains the root certicate chain from the Certicate Authority. Consider the following syntax for mongod (page 1021):
mongod --sslOnNormalPorts --sslPEMKeyFile <pem> --sslCAFile <ca>
For example, given a signed SSL certicate located at http://docs.mongodb.org/manual/etc/ssl/mongodb.pem and the certicate authority le at http://docs.mongodb.org/manual/etc/ssl/ca.pem, you can congure mongod (page 1021) for SSL encryption as follows:
mongod --sslOnNormalPorts --sslPEMKeyFile /etc/ssl/mongodb.pem --sslCAFile /etc/ssl/ca.pem
Note: Specify the <pem> le and the <ca> le with either the full path name or the relative path name. If the <pem> is encrypted, specify the encryption password with the sslPEMKeyPassword (page 1091) option. You may also specify these options in the conguration le (page 1078), as in the following example:
sslOnNormalPorts = true sslPEMKeyFile = /etc/ssl/mongodb.pem sslCAFile = /etc/ssl/ca.pem
To connect, to mongod (page 1021) and mongos (page 1032) instances using SSL, the mongo (page 1036) tools must include the both the --ssl (page 1037) and --sslPEMKeyFile (page 1037) option. See SSL Conguration
76
Chapter 7. Connect to MongoDB with SSL
for Clients (page 77) for more information on connecting to mongod (page 1021) and mongos (page 1032) running with SSL. Block Revoked Certicates for Clients To prevent clients with revoked certicates from connecting, include the sslCRLFile (page 1092) to specify a .pem le that contains revoked certicates. For example, the following mongod (page 1021) with SSL conguration includes the sslCRLFile (page 1092) setting:
mongod --sslOnNormalPorts --sslCRLFile /etc/ssl/ca-crl.pem --sslPEMKeyFile /etc/ssl/mongodb.pem --ssl
Clients with revoked certicates in the http://docs.mongodb.org/manual/etc/ssl/ca-crl.pem will not be able to connect to this mongod (page 1021) instance. Validate Only if a Client Presents a Certicate In most cases it is important to ensure that clients present valid certicates. However, if you have clients that cannot present a client certicate, or are transitioning to using a certicate authority you may only want to validate certicates from clients that present a certicate. If you want to bypass validation for clients that dont present certicates, include the sslWeakCertificateValidation (page 1092) run-time option with mongod (page 1021) and mongos (page 1032). If the client does not present a certicate, no validation occurs. These connections, though not validated, are still encrypted using SSL. For example, consider the following mongod (page 1021) with an SSL conguration that includes the sslWeakCertificateValidation (page 1092) setting:
mongod --sslOnNormalPorts --sslWeakCertificateValidation --sslPEMKeyFile /etc/ssl/mongodb.pem --sslCA
Then, clients can connect either with the option --ssl (page 1037) and no certicate or with the option --ssl (page 1037) and a valid certicate. See SSL Conguration for Clients (page 77) for more information on SSL connections for clients. Note: If the client presents a certicate, the certicate must be a valid certicate. All connections, including those that have not presented certicates are encrypted using SSL.
7.1.4 Run in FIPS Mode

If your mongod (page 1021) or mongos (page 1032) is running on a system with an OpenSSL library congured with the FIPS 140-2 module, you can run mongod (page 1021) or mongos (page 1032) in FIPS mode, with the sslFIPSMode (page 1092) setting.
7.2 SSL Conguration for Clients

Clients must have support for SSL to work with a mongod (page 1021) or a mongos (page 1032) instance that has SSL support enabled. The current versions of the Python, Java, Ruby, Node.js, .NET, and C++ drivers have support for SSL, with full support coming in future releases of other drivers.
7.2. SSL Conguration for Clients
77
7.2.1 mongo SSL Conguration

For SSL connections, you must use the mongo (page 1036) shell built with SSL support or distributed with MongoDB Enterprise. To support SSL, mongo (page 1036) has the following settings: --ssl (page 1037) --sslPEMKeyFile (page 1091) with the name of the .pem le that contains the SSL certicate and key. --sslCAFile (page 1091) with the name of the .pem le that contains the certicate from the Certicate Authority. --sslPEMKeyPassword (page 1091) option if the client certicate-key le is encrypted. Connect to MongoDB Instance with SSL Encryption To connect to a mongod (page 1021) or mongos (page 1032) instance that requires only a SSL encryption mode (page 75), start mongo (page 1036) shell with --ssl (page 1037), as in the following:
mongo --ssl
Connect to MongoDB Instance that Requires Client Certicates To connect to a mongod (page 1021) or mongos (page 1032) that requires CA-signed client certicates (page 76), start the mongo (page 1036) shell with --ssl (page 1037) and the --sslPEMKeyFile (page 1091) option to specify the signed certicate-key le, as in the following:
mongo --ssl --sslPEMKeyFile /etc/ssl/client.pem
Connect to MongoDB Instance that Validates when Presented with a Certicate To connect to a mongod (page 1021) or mongos (page 1032) instance that only requires valid certicates when the client presents a certicate (page 77), start mongo (page 1036) shell either with the --ssl (page 1037) ssl and no certicate or with the --ssl (page 1037) ssl and a valid signed certicate. For example, if mongod (page 1021) is running with weak certicate validation, both of the following mongo (page 1036) shell clients can connect to that mongod (page 1021):
mongo --ssl mongo --ssl --sslPEMKeyFile /etc/ssl/client.pem
Important: If the client presents a certicate, the certicate must be valid.
7.2.2 MMS
The MMS agent will also have to connect via SSL in order to gather its stats. Because the agent already utilizes SSL for its communications to the MMS servers, this is just a matter of enabling SSL support in MMS itself on a per host basis. Use the Edit host button (i.e. the pencil) on the Hosts page in the MMS console and is currently enabled on a group by group basis by 10gen. Please see the MMS Manual for more information about MMS conguration.
78
7.2.3 PyMongo
Add the ssl=True parameter to a PyMongo MongoClient to create a MongoDB connection to an SSL MongoDB instance:
from pymongo import MongoClient c = MongoClient(host="mongodb.example.net", port=27017, ssl=True)
To connect to a replica set, use the following operation:

from pymongo import MongoReplicaSetClient c = MongoReplicaSetClient("mongodb.example.net:27017", replicaSet="mysetname", ssl=True)
PyMongo also supports an ssl=true option for the MongoDB URI:

mongodb://mongodb.example.net:27017/?ssl=true
7.2.4 Java
Consider the following example SSLApp.java class le:
import com.mongodb.*; import javax.net.ssl.SSLSocketFactory; public class SSLApp { public static void main(String args[]) throws Exception {
MongoClientOptions o = new MongoClientOptions.Builder() .socketFactory(SSLSocketFactory.getDefault()) .build(); MongoClient m = new MongoClient("localhost", o); DB db = m.getDB( "test" ); DBCollection c = db.getCollection( "foo" ); System.out.println( c.findOne() ); } }
7.2.5 Ruby
The recent versions of the Ruby driver have support for connections to SSL servers. Install the latest version of the driver with the following command:
gem install mongo
Then connect to a standalone instance, using the following form:

require rubygems require mongo connection = MongoClient.new(localhost, 27017, :ssl => true)
Replace connection with the following if youre connecting to a replica set:
7.2. SSL Conguration for Clients
79
connection = MongoReplicaSetClient.new([localhost:27017], [localhost:27018], :ssl => true)
Here, mongod (page 1021) instance run on localhost:27017 and localhost:27018.
7.2.6 Node.JS (node-mongodb-native)

In the node-mongodb-native driver, use the following invocation to connect to a mongod (page 1021) or mongos (page 1032) instance via SSL:
var db1 = new Db(MONGODB, new Server("127.0.0.1", 27017, { auto_reconnect: false, poolSize:4, ssl:ssl } );
To connect to a replica set via SSL, use the following form:

var replSet = new ReplSetServers( [ new Server( RS.host, RS.ports[1], { auto_reconnect: true } ), new Server( RS.host, RS.ports[0], { auto_reconnect: true } ), ], {rs_name:RS.name, ssl:ssl} );
7.2.7 .NET
As of release 1.6, the .NET driver supports SSL connections with mongod (page 1021) and mongos (page 1032) instances. To connect using SSL, you must add an option to the connection string, specifying ssl=true as follows:
var connectionString = "mongodb://localhost/?ssl=true"; var server = MongoServer.Create(connectionString);
The .NET driver will validate the certicate against the local trusted certicate store, in addition to providing encryption of the server. This behavior may produce issues during testing if the server uses a self-signed certicate. If you encounter this issue, add the sslverifycertificate=false option to the connection string to prevent the .NET driver from validating the certicate, as follows:
var connectionString = "mongodb://localhost/?ssl=true&sslverifycertificate=false"; var server = MongoServer.Create(connectionString);
80
CHAPTER 8
Monitor MongoDB with SNMP
New in version 2.2. Enterprise Feature This feature is only available in MongoDB Enterprise. This document outlines the use and operation of MongoDBs SNMP extension, which is only available in MongoDB Enterprise.
8.1 Prerequisites
8.1.1 Install MongoDB Enterprise
MongoDB Enterprise
8.1.2 Included Files

The Enterprise packages contain the following les: MONGO-MIB.txt: The MIB le that describes the data (i.e. schema) for MongoDBs SNMP output mongod.conf: The SNMP conguration le for reading the SNMP output of MongoDB. The SNMP congures the community names, permissions, access controls, etc.
8.1.3 Required Packages

To use SNMP, you must install several prerequisites. The names of the packages vary by distribution and are as follows: Ubuntu 11.04 requires libssl0.9.8, snmp-mibs-downloader, snmp, and snmpd. Issue a command such as the following to install these packages:
sudo apt-get install libssl0.9.8 snmp snmpd snmp-mibs-downloader
81
Red Hat Enterprise Linux 6.x series and Amazon Linux AMI require libssl, net-snmp, net-snmp-libs, and net-snmp-utils. Issue a command such as the following to install these packages:
sudo yum install libssl net-snmp net-snmp-libs net-snmp-utils
SUSE Enterprise Linux requires libopenssl0_9_8, libsnmp15, slessp1-libsnmp15, and snmp-mibs. Issue a command such as the following to install these packages:
sudo zypper install libopenssl0_9_8 libsnmp15 slessp1-libsnmp15 snmp-mibs
8.2 Congure SNMP

8.2.1 Install MIB Conguration Files
Ensure that the MIB directory http://docs.mongodb.org/manual/usr/share/snmp/mibs exists. If not, issue the following command:
sudo mkdir -p /usr/share/snmp/mibs
Use the following command to create a symbolic link:

sudo ln -s [/path/to/mongodb/distribution/]MONGO-MIB.txt /usr/share/snmp/mibs/
Replace [/path/to/mongodb/distribution/] with the path to your MONGO-MIB.txt conguration le. Copy the mongod.conf le into the http://docs.mongodb.org/manual/etc/snmp directory with the following command:
cp mongod.conf /etc/snmp/mongod.conf
8.2.2 Start Up
You can control MongoDB Enterprise using default or custom control scripts, just as with any other mongod: Use the following command to view all SNMP options available in your MongoDB:
mongod --help | grep snmp
The above command should return the following output:

Module snmp options: --snmp-subagent --snmp-master run snmp subagent run snmp as master
Ensure that the following directories exist: http://docs.mongodb.org/manual/data/db/ (This is the path where MongoDB stores the data les.) http://docs.mongodb.org/manual/var/log/mongodb/ (This is the path where MongoDB writes the log output.) If they do not, issue the following command:
mkdir -p /var/log/mongodb/ /data/db/
82
Chapter 8. Monitor MongoDB with SNMP
Start the mongod instance with the following command:

mongod --snmp-master --port 3001 --fork --dbpath /data/db/ --logpath /var/log/mongodb/1.log
Optionally, you can set these options in a conguration le (page 1078). To check if mongod is running with SNMP support, issue the following command:
ps -ef | grep mongod --snmp
The command should return output that includes the following line. This indicates that the proper mongod instance is running:
systemuser 31415 10260 0 Jul13 pts/16 00:00:00 mongod --snmp-master --port 3001 # [...]
8.2.3 Test SNMP

Check for the snmp agent process listening on port 1161 with the following command:
sudo lsof -i :1161
which return the following output:

COMMAND PID USER mongod 9238 sysadmin FD 10u TYPE DEVICE SIZE/OFF NODE NAME IPv4 96469 0t0 UDP localhost:health-polling
Similarly, this command:

netstat -an | grep 1161
should return the following output:

udp 0 0 127.0.0.1:1161 0.0.0.0:*
8.2.4 Run snmpwalk Locally

snmpwalk provides tools for retrieving and parsing the SNMP data according to the MIB. If you installed all of the required packages above, your system will have snmpwalk. Issue the following command to collect data from mongod using SNMP:
snmpwalk -m MONGO-MIB -v 2c -c mongodb 127.0.0.1:1161 1.3.6.1.4.1.37601
You may also choose to specify the path to the MIB le:
snmpwalk -m /usr/share/snmp/mibs/MONGO-MIB -v 2c -c mongodb 127.0.0.1:1161 1.3.6.1.4.1.37601
Use this command only to ensure that you can retrieve and validate SNMP data from MongoDB.
8.3 Troubleshooting
Always check the logs for errors if something does not run as expected; see the log at http://docs.mongodb.org/manual/var/log/mongodb/1.log. The presence of the following line indicates that the mongod cannot read the http://docs.mongodb.org/manual/etc/snmp/mongod.conf le:
8.3. Troubleshooting
83
[SNMPAgent] warning: error starting SNMPAgent as master err:1
84
Chapter 8. Monitor MongoDB with SNMP
CHAPTER 9
Manage mongod Processes
MongoDB runs as a standard program. You can start MongoDB from a command line by issuing the mongod (page 1021) command and specifying options. For a list of options, see mongod (page 1021). MongoDB can also run as a Windows service. For details, see MongoDB as a Windows Service (page 18). To install MongoDB, see Install MongoDB (page 3). The following examples assume the directory containing the mongod (page 1021) process is in your system paths. The mongod (page 1021) process is the primary database process that runs on an individual server. mongos (page 1032) provides a coherent MongoDB interface equivalent to a mongod (page 1021) from the perspective of a client. The mongo (page 1036) binary provides the administrative shell. This document page discusses the mongod (page 1021) process; however, some portions of this document may be applicable to mongos (page 1032) instances. See also: Run-time Database Conguration (page 35), mongod (page 1021), mongos (page 1032), and Conguration File Options (page 1078).
9.1 Start mongod

By default, MongoDB stores data in the http://docs.mongodb.org/manual/data/db directory. On Windows, MongoDB stores data in C:\data\db. On all platforms, MongoDB listens for connections from clients on port 27017. To start MongoDB using all defaults, issue the following command at the system shell:
mongod
9.1.1 Specify a Data Directory

If you want mongod (page 1021) to store data les at a path other than http://docs.mongodb.org/manual/data/db you can specify a dbpath (page 1081). The dbpath (page 1081) must exist before you start mongod (page 1021). If it does not exist, create the directory and the permissions so that mongod (page 1021) can read and write data to this path. For more information on permissions, see the security operations documentation (page 134). To specify a dbpath (page 1081) for mongod (page 1021) to use as a data directory, use the --dbpath (page 1023) option. The following invocation will start a mongod (page 1021) instance and store data in the http://docs.mongodb.org/manual/srv/mongodb path
85
mongod --dbpath /srv/mongodb/
9.1.2 Specify a TCP Port

Only a single process can listen for connections on a network interface at a time. If you run multiple mongod (page 1021) processes on a single machine, or have other processes that must use this port, you must assign each a different port to listen on for client connections. To specify a port to mongod (page 1021), use the --port (page 1022) option on the command line. The following command starts mongod (page 1021) listening on port 12345:
mongod --port 12345
Use the default port number when possible, to avoid confusion.
9.1.3 Start mongod as a Daemon

To run a mongod (page 1021) process as a daemon (i.e. fork (page 1081),) and write its output to a log le, use the --fork (page 1023) and --logpath (page 1022) options. You must create the log directory; however, mongod (page 1021) will create the log le if it does not exist. The following command starts mongod (page 1021) as a daemon and records log output to http://docs.mongodb.org/manual/var/log/mongodb.log.
mongod --fork --logpath /var/log/mongodb.log
9.1.4 Additional Conguration Options

For an overview of common congurations and common conguration deployments. congurations for common use cases, see Run-time Database Conguration (page 35).
9.2 Stop mongod

To stop a mongod (page 1021) instance not running as a daemon, press Control+C. MongoDB stops when all ongoing operations are complete and does a clean exit, ushing and closing all data les. To stop a mongod (page 1021) instance running in the background or foreground, issue the shutdownServer() (page 985) helper in the mongo (page 1036) shell. Use the following sequence: 1. To open the mongo (page 1036) shell for a mongod (page 1021) instance running on the default port of 27017, issue the following command:
mongo
2. To switch to the admin database and shutdown the mongod (page 1021) instance, issue the following commands:
use admin db.shutdownServer()
You may only use db.shutdownServer() (page 985) when connected to the mongod (page 1021) when authenticated to the admin database or on systems without authentication connected via the localhost interface. Alternately, you can shut down the mongod (page 1021) instance: 86 Chapter 9. Manage mongod Processes
using the --shutdown (page 1028) option from a driver using the shutdown (page 869). For details, see the drivers documentation (page 555) for your driver.
9.2.1 mongod Shutdown and Replica Sets

If the mongod (page 1021) is the primary in a replica set, the shutdown process for these mongod (page 1021) instances has the following steps: 1. Check how up-to-date the secondaries are. 2. If no secondary is within 10 seconds of the primary, mongod (page 1021) will return a message that it will not shut down. You can pass the shutdown (page 869) command a timeoutSecs argument to wait for a secondary to catch up. 3. If there is a secondary within 10 seconds of the primary, the primary will step down and wait for the secondary to catch up. 4. After 60 seconds or once the secondary has caught up, the primary will shut down. If there is no up-to-date secondary and you want the primary to shut down, issue the shutdown (page 869) command with the force argument, as in the following mongo (page 1036) shell operation:
db.adminCommand({shutdown : 1, force : true})
To keep checking the secondaries for a specied number of seconds if none are immediately up-to-date, issue shutdown (page 869) with the timeoutSecs argument. MongoDB will keep checking the secondaries for the specied number of seconds if none are immediately up-to-date. If any of the secondaries catch up within the allotted time, the primary will shut down. If no secondaries catch up, it will not shut down. The following command issues shutdown (page 869) with timeoutSecs set to 5:
db.adminCommand({shutdown : 1, timeoutSecs : 5})
Alternately you can use the timeoutSecs argument with the shutdownServer() (page 985) method:
db.shutdownServer({timeoutSecs : 5})
9.3 Sending a UNIX INT or TERM Signal

You can cleanly stop mongod (page 1021) using a SIGINT or SIGTERM signal on UNIX-like systems. Either ^C for a non-daemon mongod (page 1021) instance, kill -2 <pid>, or kill -15 <pid> will cleanly terminate the mongod (page 1021) instance. Terminating a mongod (page 1021) instance that is not running with journaling with kill -9 <pid> (i.e. SIGKILL) will probably cause data corruption. To recover data in situations where mongod (page 1021) instances have not terminated cleanly without journaling see Recover MongoDB Data following Unexpected Shutdown (page 52).
9.3. Sending a UNIX INT or TERM Signal
87
88
Chapter 9. Manage mongod Processes
CHAPTER 10
Rotate Log Files
10.1 Overview
Log rotation archives the current log le and starts a new one. Specically, log rotation renames the current log le by appending the lename with a timestamp, 1 opens a new log le, and nally closes the old log. MongoDB will only rotate logs, when you use the logRotate (page 866) command, or issue the process a SIGUSR1 signal as described in this procedure. See also: For information on logging, see the Process Logging (page 94) section.
10.2 Procedure
The following steps create and rotate a log le: 1. Start a mongod (page 1021) with verbose logging, with appending enabled, and with the following log le:
mongod -v --logpath /var/log/mongodb/server1.log --logappend
2. In a separate terminal, list the matching les:

ls /var/log/mongodb/server1.log*
For results, you get:

server1.log
3. Rotate the log le using one of the following methods. From the mongo (page 1036) shell, issue the logRotate (page 866) command from the admin database:
use admin db.runCommand( { logRotate : 1 } )
This is the only available method to rotate log les on Windows systems. From the UNIX shell, rotate logs for a single process by issuing the following command:
1
MongoDB renders this timestamp in UTC (GMT) and formatted as ISODate.
89
kill -SIGUSR1 <mongod process id>
From the UNIX shell, rotate logs for all mongod (page 1021) processes on a machine by issuing the following command:
killall -SIGUSR1 mongod
4. List the matching les again:

ls /var/log/mongodb/server1.log*
For results you get something similar to the following. The timestamps will be different.
server1.log server1.log.2011-11-24T23-30-00
The example results indicate a log rotation performed at exactly 11:30 pm on November 24th, 2011 UTC, which is the local time offset by the local time zone. The original log le is the one with the timestamp. The new log is server1.log le. If you issue a second logRotate (page 866) command an hour later, then an additional le would appear when listing matching les, as in the following example:
server1.log server1.log.2011-11-24T23-30-00 server1.log.2011-11-25T00-30-00
This operation does not modify the server1.log.2011-11-24T23-30-00 le created earlier, while server1.log.2011-11-25T00-30-00 is the previous server1.log le, renamed. server1.log is a new, empty le that receives all new log output.
90
Chapter 10. Rotate Log Files
CHAPTER 11
Monitoring for MongoDB
Monitoring is a critical component of all database administration. A rm grasp of MongoDBs reporting will allow you to assess the state of your database and maintain your deployment without crisis. Additionally, a sense of MongoDBs normal operational parameters will allow you to diagnose issues as you encounter them, rather than waiting for a crisis or failure. This document provides an overview of the available tools and data provided by MongoDB as well as an introduction to diagnostic strategies, and suggestions for monitoring instances in MongoDBs replica sets and sharded clusters. Note: 10gen provides a hosted monitoring service which collects and aggregates these data to provide insight into the performance and operation of MongoDB deployments. See the MongoDB Monitoring Service (MMS) and the MMS documentation for more information.
11.1 Monitoring Tools

There are two primary methods for collecting data regarding the state of a running MongoDB instance. First, there are a set of tools distributed with MongoDB that provide real-time reporting of activity on the database. Second, several database commands (page 807) return statistics regarding the current database state with greater delity. Both methods allow you to collect data that answers a different set of questions, and are useful in different contexts. This section provides an overview of these utilities and statistics, along with an example of the kinds of questions that each method is most suited to help you address.
11.1.1 Utilities
The MongoDB distribution includes a number of utilities that return statistics about instances performance and activity quickly. These are typically most useful for diagnosing issues and assessing normal operation. mongotop mongotop (page 1068) tracks and reports the current read and write activity of a MongoDB instance. mongotop (page 1068) provides per-collection visibility into use. Use mongotop (page 1068) to verify that activity and use match expectations. See the mongotop manual (page 1067) for details.
91
mongostat mongostat (page 1063) captures and returns counters of database operations. mongostat (page 1063) reports operations on a per-type (e.g. insert, query, update, delete, etc.) basis. This format makes it easy to understand the distribution of load on the server. Use mongostat (page 1063) to understand the distribution of operation types and to inform capacity planning. See the mongostat manual (page 1063) for details. REST Interface MongoDB provides a REST interface that exposes a diagnostic and monitoring information in a simple web page. Enable this by setting rest (page 1085) to true, and access this page via the local host interface using the port numbered 1000 more than that the database port. In default congurations the REST interface is accessible on 28017. For example, to access the REST interface on a locally running mongod instance: http://localhost:28017
11.1.2 Statistics
MongoDB provides a number of commands that return statistics about the state of the MongoDB instance. These data may provide ner granularity regarding the state of the MongoDB instance than the tools above. Consider using their output in scripts and programs to develop custom alerts, or to modify the behavior of your application in response to the activity of your instance. serverStatus Access serverStatus data (page 889) by way of the serverStatus (page 889) command. This document contains a general overview of the state of the database, including disk usage, memory use, connection, journaling, index accesses. The command returns quickly and does not impact MongoDB performance. While this output contains a (nearly) complete account of the state of a MongoDB instance, in most cases you will not run this command directly. Nevertheless, all administrators should be familiar with the data provided by serverStatus (page 889). See also: db.serverStatus() (page 984) and serverStatus data (page 889). replSetGetStatus View the replSetGetStatus data (page 840) with the replSetGetStatus (page 840) command (rs.status() (page 990) from the shell). The document returned by this command reects the state and conguration of the replica set. Use this data to ensure that replication is properly congured, and to check the connections between the current host and the members of the replica set. dbStats The dbStats data (page 877) is accessible by way of the dbStats (page 877) command (db.stats() (page 986) from the shell). This command returns a document that contains data that reects the amount of storage used and data contained in the database, as well as object, collection, and index counters. Use this data to check and track the state and storage of a specic database. This output also allows you to compare utilization between databases and to determine average document size in a database.
92
Chapter 11. Monitoring for MongoDB
collStats The collStats data (page 873) is accessible using the collStats (page 873) command (db.printCollectionStats() (page 982) from the shell). It provides statistics that resemble dbStats (page 877) on the collection level: this includes a count of the objects in the collection, the size of the collection, the amount of disk space used by the collection, and information about the indexes.
11.1.3 Introspection Tools

In addition to status reporting, MongoDB provides a number of introspection tools that you can use to diagnose and analyze performance and operational conditions. Consider the following documentation: diagLogging (page 879) Analyze Performance of Database Operations (page 99) Database Proler Output (page 1100) db.currentOp() (page 971)
11.1.4 Third Party Tools

A number of third party monitoring tools have support for MongoDB, either directly, or through their own plugins. Self Hosted Monitoring Tools These are monitoring tools that you must install, congure and maintain on your own servers, usually open source. Tool Ganglia Ganglia Motop mtop Munin Munin Description Python script to report operations per second, memory usage, btree statistics, master/slave status and current connections. gmond_python_modules Parses output from the serverStatus (page 889) and replSetGetStatus (page 840) commands. None Realtime monitoring tool for several MongoDB servers. Shows current operations ordered by durations every second. None A top like tool. mongo-munin Retrieves server statistics. mongomon Retrieves collection statistics (sizes, index sizes, and each (congured) collection count for one DB). Munin munin-plugins Some additional munin plugins not in the main distribution. Ubuntu PPA Nanagios-pluginA simple Nagios check script, written in Python. gios mongodb ZabmikoomiMonitors availability, resource utilization, health, performance and other important bix mongodb metrics. Also consider dex, an index and query analyzing tool for MongoDB that compares MongoDB log les and indexes to make indexing recommendations. Hosted (SaaS) Monitoring Tools These are monitoring tools provided as a hosted service, usually on a subscription billing basis. Plugin mongodb-ganglia
11.1. Monitoring Tools
93
Name Scout Server Density
Notes Several plugins including: MongoDB Monitoring, MongoDB Slow Queries and MongoDB Replica Set Monitoring. Dashboard for MongoDB, MongoDB specic alerts, replication failover timeline and iPhone, iPad and Android mobile apps.
11.2 Process Logging

During normal operation, mongod (page 1021) and mongos (page 1032) instances report information that reect current operation to standard output, or a log le. The following runtime settings control these options. quiet (page 1086). Limits the amount of information written to the log or output. verbose (page 1079). Increases the amount of information written to the log or output. You can also specify this as v (as in -v.) Set multiple v, as in vvvv = True for higher levels of verbosity. You can also change the verbosity of a running mongod (page 1021) or mongos (page 1032) instance with the setParameter (page 869) command. logpath (page 1080). Enables logging to a le, rather than standard output. Specify the full path to the log le to this setting. logappend (page 1080). Adds information to a log le instead of overwriting the le. Note: You can specify these conguration operations as the command line arguments to mongod (page 1021) or mongos (page 1032) Additionally, the following database commands affect logging: getLog (page 880). Displays recent messages from the mongod (page 1021) process log. logRotate (page 866). Rotates the log les for mongod (page 1021) processes only. See Rotate Log Files (page 89).
11.3 Diagnosing Performance Issues

Degraded performance in MongoDB can be the result of an array of causes, and is typically a function of the relationship among the quantity of data stored in the database, the amount of system RAM, the number of connections to the database, and the amount of time the database spends in a lock state. In some cases performance issues may be transient and related to trafc load, data access patterns, or the availability of hardware on the host system for virtualized environments. Some users also experience performance limitations as a result of inadequate or inappropriate indexing strategies, or as a consequence of poor schema design patterns. In other situations, performance issues may indicate that the database may be operating at capacity and that it is time to add additional capacity to the database.
11.3.1 Locks
MongoDB uses a locking system to ensure consistency. However, if certain operations are long-running, or a queue forms, performance slows as requests and operations wait for the lock. Because lock related slow downs can be intermittent, look to the data in the globalLock (page 892) section of the serverStatus (page 889) response to assess if the lock has been a challenge to your performance. If globalLock.currentQueue.total (page 893) is consistently high, then there is a chance that a large number of requests are waiting for a lock. This indicates a possible concurrency issue that might affect performance. 94 Chapter 11. Monitoring for MongoDB
If globalLock.totalTime (page 892) is high in context of uptime (page 890) then the database has existed in a lock state for a signicant amount of time. If globalLock.ratio (page 893) is also high, MongoDB has likely been processing a large number of long running queries. Long queries are often the result of a number of factors: ineffective use of indexes, non-optimal schema design, poor query structure, system architecture issues, or insufcient RAM resulting in page faults (page 95) and disk reads.
11.3.2 Memory Usage

Because MongoDB uses memory mapped les to store data, given a data set of sufcient size, the MongoDB process will allocate all memory available on the system for its use. Because of the way operating systems function, the amount of allocated RAM is not a useful reection of MongoDBs state. While this is part of the design, and affords MongoDB superior performance, the memory mapped les make it difcult to determine if the amount of RAM is sufcient for the data set. Consider memory usage statuses (page 894) to better understand MongoDBs memory utilization. Check the resident memory use (i.e. mem.resident (page 894):) if this exceeds the amount of system memory and theres a signicant amount of data on disk that isnt in RAM, you may have exceeded the capacity of your system. Also check the amount of mapped memory (i.e. mem.mapped (page 894).) If this value is greater than the amount of system memory, some operations will require disk access page faults to read data from virtual memory with deleterious effects on performance.
11.3.3 Page Faults

Page faults represent the number of times that MongoDB requires data not located in physical memory, and must read from virtual memory. To check for page faults, see the extra_info.page_faults (page 895) value in the serverStatus (page 889) command. This data is only available on Linux systems. Alone, page faults are minor and complete quickly; however, in aggregate, large numbers of page fault typically indicate that MongoDB is reading too much data from disk and can indicate a number of underlying causes and recommendations. In many situations, MongoDBs read locks will yield after a page fault to allow other processes to read and avoid blocking while waiting for the next page to read into memory. This approach improves concurrency, and in high volume systems this also improves overall throughput. If possible, increasing the amount of RAM accessible to MongoDB may help reduce the number of page faults. If this is not possible, you may want to consider deploying a sharded cluster and/or adding one or more shards to your deployment to distribute load among mongod (page 1021) instances.
11.3.4 Number of Connections

In some cases, the number of connections between the application layer (i.e. clients) and the database can overwhelm the ability of the server to handle requests which can produce performance irregularities. Check the following elds in the serverStatus (page 889) document: globalLock.activeClients (page 893) contains a counter of the total number of clients with active operations in progress or queued. connections (page 894) is a container for the following two elds: current (page 894) the total number of current clients that connect to the database instance. available (page 894) the total number of unused collections available for new clients. Note: Unless limited by system-wide limits MongoDB has a hard connection limit of 20 thousand connections. You can modify system limits using the ulimit command, or by editing your systems http://docs.mongodb.org/manual/etc/sysctl le. 11.3. Diagnosing Performance Issues 95
If requests are high because there are many concurrent application requests, the database may have trouble keeping up with demand. If this is the case, then you will need to increase the capacity of your deployment. For readheavy applications increase the size of your replica set and distribute read operations to secondary members. For write heavy applications, deploy sharding and add one or more shards to a sharded cluster to distribute load among mongod (page 1021) instances. Spikes in the number of connections can also be the result of application or driver errors. All of the MongoDB drivers supported by 10gen implement connection pooling, which allows clients to use and reuse connections more efciently. Extremely high numbers of connections, particularly without corresponding workload is often indicative of a driver or other conguration error.
11.3.5 Database Proling

MongoDB contains a database proling system that can help identify inefcient queries and operations. Enable the proler by setting the profile (page 888) value using the following command in the mongo (page 1036) shell:
See also: The documentation of db.setProfilingLevel() (page 985) for more information about this command. Note: Because the database proler can have an impact on the performance, only enable proling for strategic intervals and as minimally as possible on production systems. You may enable proling on a per-mongod (page 1021) basis. This setting will not propagate across a replica set or sharded cluster. The following proling levels are available: Level 0 1 2 Setting Off. No proling. On. Only includes slow operations. On. Includes all operations.
See the output of the proler in the system.profile collection of your database. You can specify the slowms (page 1085) setting to set a threshold above which the proler considers operations slow and thus included in the level 1 proling data. You may congure slowms (page 1085) at runtime, as an argument to the db.setProfilingLevel() (page 985) operation. Additionally, mongod (page 1021) records all slow queries to its log (page 1080), as dened by slowms (page 1085). The data in system.profile does not persist between mongod (page 1021) restarts. You can view the prolers output by issuing the show profile command in the mongo (page 1036) shell, with the following operation.
db.system.profile.find( { millis : { $gt : 100 } } )
This returns all operations that lasted longer than 100 milliseconds. Ensure that the value specied here (i.e. 100) is above the slowms (page 1085) threshold. See also: Optimization Strategies for MongoDB (page 555) addresses strategies that may improve the performance of your database queries and operations.
96
11.4 Replication and Monitoring

The primary administrative concern that requires monitoring with replica sets, beyond the requirements for any MongoDB instance, is replication lag. This refers to the amount of time that it takes a write operation on the primary to replicate to a secondary. Some very small delay period may be acceptable; however, as replication lag grows, two signicant problems emerge: First, operations that have occurred in the period of lag are not replicated to one or more secondaries. If youre using replication to ensure data persistence, exceptionally long delays may impact the integrity of your data set. Second, if the replication lag exceeds the length of the operation log (oplog) then MongoDB will have to perform an initial sync on the secondary, copying all data from the primary and rebuilding all indexes. In normal circumstances this is uncommon given the typical size of the oplog, but its an issue to be aware of. For causes of replication lag, see Replication Lag (page 444). Replication issues are most often the result of network connectivity issues between members or the result of a primary that does not have the resources to support application and replication trafc. To check the status of a replica, use the replSetGetStatus (page 840) or the following helper in the shell:
rs.status()
See the replSetGetStatus (page 840) document for a more in depth overview view of this output. In general watch the value of optimeDate (page 842). Pay particular attention to the difference in time between the primary and the secondary members. The size of the operation log is only congurable during the rst run using the --oplogSize (page 1028) argument to the mongod (page 1021) command, or preferably the oplogSize (page 1088) in the MongoDB conguration le. If you do not specify this on the command line before running with the --replSet (page 1028) option, mongod (page 1021) will create a default sized oplog. By default the oplog is 5% of total available disk space on 64-bit systems. See also: Change the Size of the Oplog (page 435)
11.5 Sharding and Monitoring

In most cases the components of sharded clusters benet from the same monitoring and analysis as all other MongoDB instances. Additionally, clusters require monitoring to ensure that data is effectively distributed among nodes and that sharding operations are functioning appropriately. See also: See the Sharding (page 483) page for more information.
11.5.1 Cong Servers

The cong database provides a map of documents to shards. The cluster updates this map as chunks move between shards. When a conguration server becomes inaccessible, some sharding operations like moving chunks and starting mongos (page 1032) instances become unavailable. However, clusters remain accessible from already-running mongos (page 1032) instances. Because inaccessible conguration servers can have a serious impact on the availability of a sharded cluster, you should monitor the conguration servers to ensure that the cluster remains well balanced and that mongos (page 1032) instances can restart. 11.4. Replication and Monitoring 97
11.5.2 Balancing and Chunk Distribution

The most effective sharded cluster deployments require that chunks are evenly balanced among the shards. MongoDB has a background balancer process that distributes data such that chunks are always optimally distributed among the shards. Issue the db.printShardingStatus() (page 983) or sh.status() (page 1002) command to the mongos (page 1032) by way of the mongo (page 1036) shell. This returns an overview of the entire cluster including the database name, and a list of the chunks.
11.5.3 Stale Locks

In nearly every case, all locks used by the balancer are automatically released when they become stale. However, because any long lasting lock can block future balancing, its important to insure that all locks are legitimate. To check the lock status of the database, connect to a mongos (page 1032) instance using the mongo (page 1036) shell. Issue the following command sequence to switch to the config database and display all outstanding locks on the shard database:
use config db.locks.find()
For active deployments, the above query might return a useful result set. The balancing process, which originates on a randomly selected mongos (page 1032), takes a special balancer lock that prevents other balancing activity from transpiring. Use the following command, also to the config database, to check the status of the balancer lock.
db.locks.find( { _id : "balancer" } )
If this lock exists, make sure that the balancer process is actively using this lock.
98
CHAPTER 12
Analyze Performance of Database Operations
The database proler collects ne grained data about MongoDB write operations, cursors, database commands on a running mongod (page 1021) instance. You can enable proling on a per-database or per-instance basis. The proling level (page 99) is also congurable when enabling proling. The database proler writes all the data it collects to the system.profile (page 1099) collection, which is a capped collection (page 558). See Database Proler Output (page 1100) for overview of the data in the system.profile (page 1099) documents created by the proler. This document outlines a number of key administration options for the database proler. For additional related information, consider the following resources: Database Proler Output (page 1100) Prole Command (page 888) db.currentOp() (page 971)
12.1 Proling Levels

The following proling levels are available: 0 - the proler is off, does not collect any data. 1 - collects proling data for slow operations only. By default slow operations are those slower than 100 milliseconds. You can modify the threshold for slow operations with the slowms (page 1085) runtime option or the setParameter (page 869) command. See the Specify the Threshold for Slow Operations (page 100) section for more information. 2 - collects proling data for all database operations.
12.2 Enable Database Proling and Set the Proling Level

You can enable database proling from the mongo (page 1036) shell or through a driver using the profile (page 888) command. This section will describe how to do so from the mongo (page 1036) shell. See your driver documentation (page 555) if you want to control the proler from within your application. When you enable proling, you also set the proling level (page 99). The proler records data in the system.profile (page 1099) collection. MongoDB creates the system.profile (page 1099) collection in a database after you enable proling for that database.
99
To enable proling and set the proling level, issue use the db.setProfilingLevel() (page 985) helper in the mongo (page 1036) shell, passing the proling level as a parameter. For example, to enable proling for all database operations, consider the following operation in the mongo (page 1036) shell:
The shell returns a document showing the previous level of proling. The "ok" : operation succeeded:
{ "was" : 0, "slowms" : 100, "ok" : 1 }
1 key-value pair indicates the
To verify the new setting, see the Check Proling Level (page 100) section.
12.2.1 Specify the Threshold for Slow Operations

The threshold for slow operations applies to the entire mongod (page 1021) instance. When you change the threshold, you change it for all databases on the instance. Important: Changing the slow operation threshold for the database proler also affects the proling subsystems slow operation threshold for the entire mongod (page 1021) instance. Always set the threshold to the highest useful value. By default the slow operation threshold is 100 milliseconds. Databases with a proling level of 1 will log operations slower than 100 milliseconds. To change the threshold, pass two parameters to the db.setProfilingLevel() (page 985) helper in the mongo (page 1036) shell. The rst parameter sets the proling level for the current database, and the second sets the default slow operation threshold for the entire mongod (page 1021) instance. For example, the following command sets the proling level for the current database to 0, which disables proling, and sets the slow-operation threshold for the mongod (page 1021) instance to 20 milliseconds. Any database on the instance with a proling level of 1 will use this threshold:
db.setProfilingLevel(0,20)
12.2.2 Check Proling Level

To view the proling level (page 99), issue the following from the mongo (page 1036) shell:
db.getProfilingStatus()
The shell returns a document similar to the following:

{ "was" : 0, "slowms" : 100 }
The was eld indicates the current level of proling. The slowms eld indicates how long an operation must exist in milliseconds for an operation to pass the slow threshold. MongoDB will log operations that take longer than the threshold if the proling level is 1. This document returns the proling level in the was eld. For an explanation of proling levels, see Proling Levels (page 99). To return only the proling level, use the db.getProfilingLevel() (page 979) helper in the mongo (page 1036) as in the following:
db.getProfilingLevel()
100
Chapter 12. Analyze Performance of Database Operations
12.2.3 Disable Proling

To disable proling, use the following helper in the mongo (page 1036) shell:
12.2.4 Enable Proling for an Entire mongod Instance

For development purposes in testing environments, you can enable database proling for an entire mongod (page 1021) instance. The proling level applies to all databases provided by the mongod (page 1021) instance. To enable proling for a mongod (page 1021) instance, pass the following parameters to mongod (page 1021) at startup or within the conguration le (page 1078):
mongod --profile=1 --slowms=15
This sets the proling level to 1, which collects proling data for slow operations only, and denes slow operations as those that last longer than 15 milliseconds. See also: profile (page 1084) and slowms (page 1085).
12.2.5 Database Proling and Sharding

You cannot enable proling on a mongos (page 1032) instance. To enable proling in a shard cluster, you must enable proling for each mongod (page 1021) instance in the cluster.
12.3 View Proler Data

The database proler logs information about database operations in the system.profile (page 1099) collection. To view proling information, query the system.profile (page 1099) collection. To view example queries, see Proler Overhead (page 102) For an explanation of the output data, see Database Proler Output (page 1100).
12.3.1 Example Proler Data Queries

This section displays example queries to the system.profile (page 1099) collection. For an explanation of the query output, see Database Proler Output (page 1100). To return the most recent 10 log entries in the system.profile (page 1099) collection, run a query similar to the following:
db.system.profile.find().limit(10).sort( { ts : -1 } ).pretty()
To return all operations except command operations ($cmd), run a query similar to the following:
db.system.profile.find( { op: { $ne : command } } ).pretty()
To return operations for a particular collection, run a query similar to the following. This example returns operations in the mydb databases test collection:
12.3. View Proler Data
101
db.system.profile.find( { ns : mydb.test } ).pretty()
To return operations slower than 5 milliseconds, run a query similar to the following:
db.system.profile.find( { millis : { $gt : 5 } } ).pretty()
To return information from a certain time range, run a query similar to the following:
db.system.profile.find( { ts : { $gt : new ISODate("2012-12-09T03:00:00Z") , $lt : new ISODate("2012-12-09T03:40:00Z") } } ).pretty()
The following example looks at the time range, suppresses the user eld from the output to make it easier to read, and sorts the results by how long each operation took to run:
db.system.profile.find( { ts : { $gt : new ISODate("2011-07-12T03:00:00Z") , $lt : new ISODate("2011-07-12T03:40:00Z") } }, { user : 0 } ).sort( { millis : -1 } )
12.3.2 Show the Five Most Recent Events

On a database that has proling enabled, the show profile helper in the mongo (page 1036) shell displays the 5 most recent operations that took at least 1 millisecond to execute. Issue show profile from the mongo (page 1036) shell, as follows:
show profile
12.4 Proler Overhead

When enabled, proling has a minor effect on performance. The system.profile (page 1099) collection is a capped collection with a default size of 1 megabyte. A collection of this size can typically store store several thousand prole documents, but some application may use more or less proling data per operation. To change the size of the system.profile (page 1099) collection, you must: 1. Disable proling. 2. Drop the system.profile (page 1099) collection. 3. Create a new system.profile (page 1099) collection. 4. Re-enable proling. For example, to create a new system.profile (page 1099) collections thats 4000000 bytes, use the following sequence of operations in the mongo (page 1036) shell:
102
db.setProfilingLevel(0) db.system.profile.drop() db.createCollection( "system.profile", { capped: true, size:4000000 } ) db.setProfilingLevel(1)
12.4. Proler Overhead
103
104
CHAPTER 13
Import and Export MongoDB Data
This document provides an overview of the import and export programs included in the MongoDB distribution. These tools are useful when you want to backup or export a portion of your data without capturing the state of the entire database, or for simple data ingestion cases. For more complex data migration tasks, you may want to write your own import and export scripts using a client driver to interact with the database itself. For disaster recovery protection and routine database backup operation, use full database instance backups (page 41). Warning: Because these tools primarily operate by interacting with a running mongod (page 1021) instance, they can impact the performance of your running database. Not only do these processes create trafc for a running database instance, they also force the database to read all data through memory. When MongoDB reads infrequently used data, it can supplant more frequently accessed data, causing a deterioration in performance for the databases regular workload. mongoimport (page 1056) and mongoexport (page 1059) do not reliably preserve all rich BSON data types, because BSON is a superset of JSON . Thus, mongoimport (page 1056) and mongoexport (page 1059) cannot represent BSON data accurately in JSON . As a result data exported or imported with these tools may lose some measure of delity. See MongoDB Extended JSON (page 1113) for more information about MongoDB Extended JSON. See also: See the Backup Strategies for MongoDB Systems (page 41) document for more information on backing up MongoDB instances. Additionally, consider the following references for commands addressed in this document: mongoexport (page 1059) mongorestore (page 1047) mongodump (page 1044) If you want to transform and process data once youve imported it in MongoDB consider the documents in the Aggregation (page 253) section, including: Map-Reduce (page 311) and Aggregation Framework (page 255).
13.1 Data Type Fidelity

JSON does not have the following data types that exist in BSON documents: data_binary, data_date, data_timestamp, data_regex, data_oid and data_ref. As a result using any tool that decodes BSON documents into JSON will suffer some loss of delity.
105
If maintaining type delity is important, consider writing a data import and export system that does not force BSON documents into JSON form as part of the process. The following list of types contain examples for how MongoDB will represent how BSON documents render in JSON. data_binary
{ "$binary" : "<bindata>", "$type" : "<t>" }
<bindata> is the base64 representation of a binary string. <t> is the hexadecimal representation of a single byte indicating the data type. data_date
Date( <date> )
<date> is the JSON representation of a 64-bit signed integer for milliseconds since epoch. data_timestamp
Timestamp( <t>, <i> )
<t> is the JSON representation of a 32-bit unsigned integer for milliseconds since epoch. <i> is a 32-bit unsigned integer for the increment. data_regex
/<jRegex>/<jOptions>
<jRegex> is a string that may contain valid JSON characters and unescaped double quote (i.e. ") characters, but may not contain unescaped forward slash (i.e. http://docs.mongodb.org/manual/) characters. <jOptions> is a string that may contain only the characters g, i, m, and s. data_oid
ObjectId( "<id>" )
<id> is a 24 character hexadecimal string. These representations require that data_oid values have an associated eld named _id. data_ref
DBRef( "<name>", "<id>" )
<name> is a string of valid JSON characters. <id> is a 24 character hexadecimal string. See also: MongoDB Extended JSON (page 1113)
13.2 Data Import and Export and Backups Operations

For resilient and non-disruptive backups, use a le system or block-level disk snapshot function, such as the methods described in the Backup Strategies for MongoDB Systems (page 41) document. The tools and operations discussed provide functionality thats useful in the context of providing some kinds of backups. By contrast, use import and export tools to backup a small subset of your data or to move data to or from a 3rd party system. These backups may capture a small crucial set of data or a frequently modied section of data, for extra insurance, or for ease of access. No matter how you decide to import or export your data, consider the following guidelines: Label les so that you can identify what point in time the export or backup reects. 106 Chapter 13. Import and Export MongoDB Data
Labeling should describe the contents of the backup, and reect the subset of the data corpus, captured in the backup or export. Do not create or apply exports if the backup process itself will have an adverse effect on a production system. Make sure that they reect a consistent data state. Export or backup processes can impact data integrity (i.e. type delity) and consistency if updates continue during the backup process. Test backups and exports by restoring and importing to ensure that the backups are useful.
13.3 Human Intelligible Import/Export Formats

This section describes a process to import/export your database, or a portion thereof, to a le in a JSON or CSV format. See also: The mongoimport (page 1055) and mongoexport (page 1059) documents contain complete documentation of these tools. If you have questions about the function and parameters of these tools not covered here, please refer to these documents. If you want to simply copy a database or collection from one instance to another, consider using the copydb (page 862), clone (page 857), or cloneCollection (page 856) commands, which may be more suited to this task. The mongo (page 1036) shell provides the db.copyDatabase() (page 970) method. These tools may also be useful for importing data into a MongoDB database from third party applications.
13.3.1 Collection Export with mongoexport

With the mongoexport (page 1059) utility you can create a backup le. In the most simple invocation, the command takes the following form:
mongoexport --collection collection --out collection.json
This will export all documents in the collection named collection into the le collection.json. Without the output specication (i.e. --out collection.json (page 1062),) mongoexport (page 1059) writes output to standard output (i.e. stdout.) You can further narrow the results by supplying a query lter using the --query (page 1061) and limit results to a single database using the --db (page 1061) option. For instance:
mongoexport --db sales --collection contacts --query {"field": 1}
This command returns all documents in the sales databases contacts collection, with a eld named field with a value of 1. Enclose the query in single quotes (e.g. ) to ensure that it does not interact with your shell environment. The resulting documents will return on standard output. By default, mongoexport (page 1059) returns one JSON document per MongoDB document. Specify the --jsonArray (page 1061) argument to return the export as a single JSON array. Use the --csv (page 1061) le to return the result in CSV (comma separated values) format. If your mongod (page 1021) instance is not running, you can use the --dbpath (page 1060) option to specify the location to your MongoDB instances database les. See the following example:
mongoexport --db sales --collection contacts --dbpath /srv/MongoDB/
This reads the data les directly. This locks the data directory to prevent conicting writes. The mongod (page 1021) process must not be running or attached to these data les when you run mongoexport (page 1059) in this conguration.
13.3. Human Intelligible Import/Export Formats
107
The --host (page 1059) and --port (page 1060) options allow you to specify a non-local host to connect to capture the export. Consider the following example:
mongoexport --host mongodb1.example.net --port 37017 --username user --password pass --collection con
On any mongoexport (page 1059) command you may, as above specify username and password credentials as above.
13.3.2 Collection Import with mongoimport

To restore a backup taken with mongoexport (page 1059). Most of the arguments to mongoexport (page 1059) also exist for mongoimport (page 1056). Consider the following command:
mongoimport --collection collection --file collection.json
This imports the contents of the le collection.json into the collection named collection. If you do not specify a le with the --file (page 1058) option, mongoimport (page 1056) accepts input over standard input (e.g. stdin.) If you specify the --upsert (page 1058) option, all of mongoimport (page 1056) operations will attempt to update existing documents in the database and insert other documents. This option will cause some performance impact depending on your conguration. You can specify the database option --db (page 1057) to import these documents to a particular database. If your MongoDB instance is not running, use the --dbpath (page 1057) option to specify the location of your MongoDB instances database les. Consider using the --journal (page 1057) option to ensure that mongoimport (page 1056) records its operations in the journal. The mongod process must not be running or attached to these data les when you run mongoimport (page 1056) in this conguration. Use the --ignoreBlanks (page 1058) option to ignore blank elds. For CSV and TSV imports, this option provides the desired functionality in most cases: it avoids inserting blank elds in MongoDB documents.
108
Chapter 13. Import and Export MongoDB Data
CHAPTER 14
Linux ulimit Settings
The Linux kernel provides a system to limit and control the number of threads, connections, and open les on a perprocess and per-user basis. These limits prevent single users from using too many system resources. Sometimes, these limits, as congured by the distribution developers, are too low for MongoDB and can cause a number of issues in the course of normal MongoDB operation. Generally, MongoDB should be the only user process on a system, to prevent resource contention.
14.1 Resource Utilization

mongod (page 1021) and mongos (page 1032) each use threads and le descriptors to track connections and manage internal operations. This section outlines the general resource utilization patterns for MongoDB. Use these gures in combination with the actual information about your deployment and its use to determine ideal ulimit settings. Generally, all mongod (page 1021) and mongos (page 1032) instances, like other processes: track each incoming connection with a le descriptor and a thread. track each internal thread or pthread as a system process.
14.1.1 mongod
1 le descriptor for each data le in use by the mongod (page 1021) instance. 1 le descriptor for each journal le used by the mongod (page 1021) instance when journal (page 1083) is true. In replica sets, each mongod (page 1021) maintains a connection to all other members of the set. mongod (page 1021) uses background threads for a number of internal processes, including TTL collections (page 577), replication, and replica set health checks, which may require a small number of additional resources.
14.1.2 mongos
In addition to the threads and le descriptors for client connections, mongos (page 1032) must maintain connects to all cong servers and all shards, which includes all members of all replica sets. For mongos (page 1032), consider the following behaviors: mongos (page 1032) instances maintain a connection pool to each shard so that the mongos (page 1032) can reuse connections and quickly fulll requests without needing to create new connections.
109
You can limit the number of incoming connections using the maxConns (page 1079) run-time option. By restricting the number of incoming connections you can prevent a cascade effect where the mongos (page 1032) creates too many connections on the mongod (page 1021) instances. Note: You cannot set maxConns (page 1079) to a value higher than 20000.
14.2 Review and Set Resource Limits

14.2.1 ulimit
You can use the ulimit command at the system prompt to check system limits, as in the following example:
$ ulimit -a -t: cpu time (seconds) -f: file size (blocks) -d: data seg size (kbytes) -s: stack size (kbytes) -c: core file size (blocks) -m: resident set size (kbytes) -u: processes -n: file descriptors -l: locked-in-memory size (kb) -v: address space (kb) -x: file locks -i: pending signals -q: bytes in POSIX msg queues -e: max nice -r: max rt priority -N 15: unlimited unlimited unlimited 8192 0 unlimited 192276 21000 40000 unlimited unlimited 192276 819200 30 65 unlimited
ulimit refers to the per-user limitations for various resources. Therefore, if your mongod (page 1021) instance executes as a user that is also running multiple processes, or multiple mongod (page 1021) processes, you might see contention for these resources. Also, be aware that the processes value (i.e. -u) refers to the combined number of distinct processes and sub-process threads. You can change ulimit settings by issuing a command in the following form:
ulimit -n <value>
For many distributions of Linux you can change values by substituting the -n option for any possible value in the output of ulimit -a. See your operating system documentation for the precise procedure for changing system limits on running systems. Note: After changing the ulimit settings, you must restart the process to take advantage of the modied settings. You can use the http://docs.mongodb.org/manual/proc le system to see the current limitations on a running process. Depending on your systems conguration, and default settings, any change to system limits made using ulimit may revert following system a system restart. Check your distribution and operating system documentation for more information.
14.2.2 /proc File System
110
Chapter 14. Linux ulimit Settings
Note: This section applies only to Linux operating systems. The http://docs.mongodb.org/manual/proc le-system stores the per-process limits in the le system object located at http://docs.mongodb.org/manual/proc/<pid>/limits, where <pid> is the processs PID or process identier. You can use the following bash function to return the content of the limits object for a process or processes with a given name:
return-limits(){ for process in $@; do process_pids=ps -C $process -o pid --no-headers | cut -d " " -f 2 if [ -z $@ ]; then echo "[no $process running]" else for pid in $process_pids; do echo "[$process #$pid -- limits]" cat /proc/$pid/limits done fi done }
You can copy and paste this function into a current shell session or load it as part of a script. Call the function with one the following invocations:
return-limits mongod return-limits mongos return-limits mongod mongos
The output of the rst command may resemble the following:

[mongod #6809 -- limits] Limit Max cpu time Max file size Max data size Max stack size Max core file size Max resident set Max processes Max open files Max locked memory Max address space Max file locks Max pending signals Max msgqueue size Max nice priority Max realtime priority Max realtime timeout Soft Limit unlimited unlimited unlimited 8720000 0 unlimited 192276 1024 40960000 unlimited unlimited 192276 819200 30 65 unlimited Hard Limit unlimited unlimited unlimited unlimited unlimited unlimited 192276 4096 40960000 unlimited unlimited 192276 819200 30 65 unlimited Units seconds bytes bytes bytes bytes bytes processes files bytes bytes locks signals bytes
us
14.3 Recommended Settings

Every deployment may have unique requirements and settings; however, the following thresholds and settings are particularly important for mongod (page 1021) and mongos (page 1032) deployments: 14.3. Recommended Settings 111
-f (le size): unlimited -t (cpu time): unlimited -v (virtual memory): unlimited 1 -n (open les): 64000 -m (memory size): unlimited 1 -u (processes/threads): 32000 Always remember to restart your mongod (page 1021) and mongos (page 1032) instances after changing the ulimit settings to make sure that the settings change takes effect.
1 If you limit virtual or resident memory size on a system running MongoDB the operating system will refuse to honor additional allocation requests.
112
Chapter 14. Linux ulimit Settings
CHAPTER 15
Production Notes
This page details system congurations that affect MongoDB, especially in production.
15.1 Backups
To make backups of your MongoDB database, please refer to Backup Strategies for MongoDB Systems (page 41).
15.2 Networking
Always run MongoDB in a trusted environment, with network rules that prevent access from all unknown machines, systems, or networks. As with any sensitive system dependent on network access, your MongoDB deployment should only be accessible to specic systems that require access: application servers, monitoring services, and other MongoDB components. See documents in the Security (page 129) section for additional information, specically: Interfaces and Port Numbers (page 132) Firewalls (page 133) Congure Linux iptables Firewall for MongoDB (page 137) Congure Windows netsh Firewall for MongoDB (page 141) For Windows users, consider the Windows Server Technet Article on TCP Conguration when deploying MongoDB on Windows.
15.3 MongoDB on Linux

If you use the Linux kernel, the MongoDB user community has recommended Linux kernel 2.6.36 or later for running MongoDB in production. Because MongoDB preallocates its database les before using them and because MongoDB uses very large les on average, you should use the Ext4 and XFS le systems if using the Linux kernel: If you use the Ext4 le system, use at least version 2.6.23 of the Linux Kernel. If you use the XFS le system, use at least version 2.6.25 of the Linux Kernel. If you are using a Red Hat derived distribution, use at least version 2.6.245.el5 of the Linux Kernel.
113
For MongoDB on Linux use the following recommended congurations: Turn off atime for the storage volume with the database les. Set the le descriptor limit and the user process limit above 20,000, according to the suggestions in Linux ulimit Settings (page 109). A low ulimit will affect MongoDB when under heavy use and will produce weird errors. Do not use hugepages virtual memory pages, MongoDB performs better with normal virtual memory pages. Disable NUMA in your BIOS. If that is not possible see NUMA (page 115). Ensure that readahead settings for the block devices that store the database les are acceptable. See the Readahead (page 114) section Use NTP to synchronize time among your hosts. This is especially important in sharded clusters.
15.4 Readahead
For random access use patterns set readahead values low, for example setting readahead to a small value such as 32 (16KB) often works well.
15.5 MongoDB on Virtual Environments

The section describes considerations when running MongoDB in some of the more common virtual environments.
15.5.1 EC2
MongoDB is compatible with EC2 and requires no conguration changes specic to the environment.
15.5.2 VMWare
MongoDB is compatible with VMWare. Some in the MongoDB community have run into issues with VMWares memory overcommit feature and suggest disabling the feature. You can clone a virtual machine running MongoDB. You might use this to spin up a new virtual host that will be added as a member of a replica set. If journaling is enabled, the clone snapshot will be consistent. If not using journaling, stop mongod (page 1021), clone, and then restart.
15.5.3 OpenVZ
The MongoDB community has encountered issues running MongoDB on OpenVZ.
15.6 Disk and Storage Systems

15.6.1 Swap
Congure swap space for your systems. Having swap can prevent issues with memory contention and can prevent the OOM Killer on Linux systems from killing mongod (page 1021). Because of the way mongod (page 1021) maps memory les to memory, the operating system will never store MongoDB data in swap.
114
Chapter 15. Production Notes
15.6.2 RAID
Most MongoDB deployments should use disks backed by RAID-10. RAID-5 and RAID-6 do not typically provide sufcient performance to support a MongoDB deployment. RAID-0 provides good write performance but provides limited availability, and reduced performance on read operations, particularly using Amazons EBS volumes: as a result, avoid RAID-0 with MongoDB deployments.
15.6.3 Remote Filesystems

Some versions of NFS perform very poorly with MongoDB and NFS is not recommended for use with MongoDB. Performance problems arise when both the data les and the journal les are both hosted on NFS: you may experience better performance if you place the journal on local or iscsi volumes. If you must use NFS, add the following NFS options to your http://docs.mongodb.org/manual/etc/fstab le: bg, nolock, and noatime. Many MongoDB deployments work successfully with Amazons Elastic Block Store (EBS) volumes. There are certain intrinsic performance characteristics, with EBS volumes that users should consider.
15.7 Hardware Requirements and Limitations

MongoDB is designed specically with commodity hardware in mind and has few hardware requirements or limitations. MongoDB core components runs on little-endian hardware primarily x86/x86_64 processors. Client libraries (i.e. drivers) can run on big or little endian systems. When installing hardware for MongoDB, consider the following: As with all software, more RAM and a faster CPU clock speed are important to productivity. Because databases do not perform high amounts of computation, increasing the number cores helps but does not provide a high level of marginal return. MongoDB has good results and good price/performance with SATA SSD (Solid State Disk) and with PCI (Peripheral Component Interconnect). Commodity (SATA) spinning drives are often a good option as the speed increase for random I/O for more expensive drives is not that dramatic (only on the order of 2x), spending that money on SSDs or RAM may be more effective.
15.7.1 MongoDB on NUMA Hardware

MongoDB and NUMA, Non-Uniform Access Memory, do not work well together. When running MongoDB on NUMA hardware, disable NUMA for MongoDB and run with an interleave memory policy. NUMA can cause a number of operational problems with MongoDB, including slow performance for periods of time or high system processor usage. Note: On Linux, MongoDB version 2.0 and greater checks these settings on start up and prints a warning if the system is NUMA-based. To disable NUMA for MongoDB, use the numactl command and start mongod (page 1021) in the following manner:
numactl --interleave=all /usr/bin/local/mongod
Adjust the proc settings using the following command:
15.7. Hardware Requirements and Limitations
115
echo 0 > /proc/sys/vm/zone_reclaim_mode
To fully disable NUMA you must perform both operations. However, you can change zone_reclaim_mode without restarting mongod. For more information, see documentation on Proc/sys/vm. See the The MySQL swap insanity problem and the effects of NUMA post, which describes the effects of NUMA on databases. This blog post addresses the impact of NUMA for MySQL; however, the issues for MongoDB are similar. The post introduces NUMA its goals, and illustrates how these goals are not compatible with production databases.
15.8 Performance Monitoring

15.8.1 iostat
On Linux, use the iostat command to check if disk I/O is a bottleneck for your database. Specify a number of seconds when running iostat to avoid displaying stats covering the time since server boot. For example:
iostat -xm 2
Use the mount command to see what device your data directory (page 1081) resides on. Key elds from iostat: %util: this is the most useful eld for a quick check, it indicates what percent of the time the device/drive is in use. avgrq-sz: average request size. Smaller number for this value reect more random IO operations.
15.8.2 bwm-ng
bwm-ng is a command-line tool for monitoring network use. If you suspect a network-based bottleneck, you may use bwm-ng to begin your diagnostic process.
15.9 Production Checklist

15.9.1 64-bit Builds for Production
Always use 64-bit Builds for Production. MongoDB uses memory mapped les. See the 32-bit limitations (page 712) for more information. 32-bit builds exist to support use on development machines and also for other miscellaneous things such as replica set arbiters.
15.9.2 BSON Document Size Limit

There is a BSON Document Size (page 1105) limit at the time of this writing 16MB per document. If you have large objects, use GridFS (page 196) instead.
116
15.9.3 Set Appropriate Write Concern for Write Operations

See Write Concern (page 398) for more information.
15.9.4 Dynamic Schema

Data in MongoDB has a dynamic schema. Collections do not enforce document structure. This facilitates iterative development and polymorphism. However, collections often hold documents with highly homogeneous structures. See Data Modeling Considerations for MongoDB Applications (page 233) for more information. Some operational considerations include: the exact set of collections to be used the indexes to be used, which are created explicitly except for the _id index shard key declarations, which are explicit and quite important as it is hard to change shard keys later One very simple rule-of-thumb is not to import data from a relational database unmodied: you will generally want to roll up certain data into richer documents that use some embedding of nested documents and arrays (and/or arrays of subdocuments).
15.9.5 Updates by Default Affect Only one Document

Set the multi parameter to true to update() (page 948) multiple documents that meet the query criteria. The mongo (page 1036) shell syntax is:
db.records.update(my_query, my_update_expression, bool_upsert, bool_multi)
Set bool_multi to true when updating many documents. Otherwise only the rst matched will update.
15.9.6 Case Sensitive Strings

MongoDB strings are case sensitive. So a search for "joe" will not nd "Joe". Consider: storing data in a normalized case format, or using regular expressions ending with http://docs.mongodb.org/manual/i and/or using $toLower (page 289) or $toUpper (page 290) in the aggregation framework (page 255)
15.9.7 Type Sensitive Fields

MongoDB data which is JSON-style, specically, BSON format have several data types. Consider the following document which has a eld x with the string value "123":
{ x : "123" }
Then the following query which looks for a number value 123 will not return that document:
db.mycollection.find( { x : 123 } )
15.9. Production Checklist
117
15.9.8 Locking
Older versions of MongoDB used a global lock; use MongoDB v2.2+ for better results. See the Concurrency (page 727) page for more information.
15.9.9 Packages
Be sure you have the latest stable release if you are using a package manager. You can see what is current on the Downloads page, even if you then choose to install via a package manager.
15.9.10 Use Odd Number of Replica Set Members

Replica sets (page 385) perform consensus elections. Use either an odd number of members (e.g., three) or else use an arbiter to get up to an odd number of votes.
15.9.11 Dont disable journaling

See Journaling (page 71) for more information.
15.9.12 Keep Replica Set Members Up-to-Date

This is important as MongoDB replica sets support automatic failover. Thus you want your secondaries to be up-todate. You have a few options here: 1. Monitoring and alerts for any lagging can be done via various means. MMS shows a graph of replica set lag 2. Using getLastError (page 398) with w:majority, you will get a timeout or no return if a majority of the set is lagging. This is thus another way to guard against lag and get some reporting back of its occurrence. 3. Or, if you want to fail over manually, you can set your secondaries to priority:0 in their conguration. Then manual action would be required for a failover. This is practical for a small cluster; for a large cluster you will want automation. Additionally, see information on replica set rollbacks (page 390).
15.9.13 Additional Deployment Considerations

Pick your shard keys carefully! There is no way to modify a shard key on a collection that is already sharded. You cannot shard an existing collection over 256 gigabytes. To shard large amounts of data, create a new empty sharded collection, and ingest the data from the source collection using an application level import operation. Unique indexes are not enforced across shards except for the shard key itself. See Enforce Unique Keys for Sharded Collections (page 533). Consider pre-splitting (page 503) a sharded collection before a massive bulk import. Usually this isnt necessary but on a bulk import of size it is helpful. Use security/auth (page 131) mode if you need it. By default auth (page 1081) is not enabled and mongod (page 1021) assumes a trusted environment. You do not have fully generalized transactions (page 571). Create rich documents and read the preceding link and consider the use case often there is a good t.
118
Disable NUMA for best results. If you have NUMA enabled, mongod (page 1021) will print a warning when it starts. Avoid excessive prefetch/readahead on the lesystem. Check your prefetch settings. Note on linux the parameter is in sectors, not bytes. 32KBytes (a setting of 64 sectors) is pretty reasonable. Check ulimits (page 109) settings. Use SSD if available and economical. Spinning disks can work well but SSDs capacity for random I/O operations work well with the update model of mongod (page 1021). See Remote Filesystems (page 115) for more info. Ensure that clients keep reasonable pool sizes to avoid overloading the connection tracking capacity of a single mongod (page 1021) or mongos (page 1032) instance.
15.9. Production Checklist
119
120
CHAPTER 16
Use Database Commands
The MongoDB command interface provides access to all non CRUD database operations. Fetching server stats, initializing a replica set, and running a map-reduce job are all accomplished with commands. See Database Commands (page 807) for list of all commands sorted by function, and Database Commands (page 807) for a list of all commands sorted alphabetically.
16.1 Database Command Form

You specify a command rst by constructing a standard BSON document whose rst key is the name of the command. For example, specify the isMaster (page 838) command using the following BSON document:
{ isMaster: 1 }
16.2 Issue Commands

The mongo (page 1036) shell provides a helper method for running commands called db.runCommand() (page 984). The following operation in mongo (page 1036) runs the above command:
db.runCommand( { isMaster: 1 } )
Many drivers (page 555) provide an equivalent for the db.runCommand() (page 984) method. Internally, running commands with db.runCommand() (page 984) is equivalent to a special query against the $cmd collection. Many common commands have their own shell helpers or wrappers in the mongo (page 1036) shell and drivers, such as the db.isMaster() (page 981) method in the mongo (page 1036) JavaScript shell.
16.3 admin Database Commands

You must run some commands on the admin database. Normally, these operations resemble the followings:
use admin db.runCommand( {buildInfo: 1} )
However, theres also a command helper that automatically runs the command in the context of the admin database:
db._adminCommand( {buildInfo: 1} )
121
16.4 Command Responses

All commands return, at minimum, a document with an ok eld indicating whether the command has succeeded:
{ ok: 1 }
Failed commands return the ok eld with a value of 0.
122
Chapter 16. Use Database Commands
CHAPTER 17
MongoDB Tutorials
This page lists the tutorials available as part of the MongoDB Manual (page 1). In addition to these documents, you can refer to the introductory MongoDB Tutorial (page 21). If there is a process or pattern that you would like to see included here, please open a Jira Case.
17.1 Getting Started

Install MongoDB on Linux (page 11) Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux (page 3) Install MongoDB on Debian (page 9) Install MongoDB on Ubuntu (page 6) Install MongoDB on OS X (page 13) Install MongoDB on Windows (page 16) Getting Started with MongoDB (page 21)
17.2 Administration
17.2.1 Replica Sets
Deploy a Replica Set (page 419) Convert a Standalone to a Replica Set (page 423) Add Members to a Replica Set (page 424) Remove Members from Replica Set (page 427) Replace a Replica Set Member (page 427) Adjust Priority for Replica Set Member (page 428) Resync a Member of a Replica Set (page 428) Deploy a Geographically Distributed Replica Set (page 429) Change the Size of the Oplog (page 435) Force a Member to Become Primary (page 437)
123
Change Hostnames in a Replica Set (page 440) Add an Arbiter to Replica Set (page 448) Convert a Secondary to an Arbiter (page 449) Congure a Secondarys Sync Target (page 454) Congure a Delayed Replica Set Member (page 451) Congure a Replica Set Member as Hidden (page 452) Congure a Non-Voting Replica Set Member (page 452) Prevent Replica Set Member from Becoming Primary (page 453) Congure Replica Set Tag Sets (page 455) Manage Chained Replication (page 439) Recongure a Replica Set with Unavailable Members (page 458) Recover MongoDB Data following Unexpected Shutdown (page 52) Troubleshoot Replica Sets (page 444)
17.2.2 Sharding
Deploy a Sharded Cluster (page 503) Convert a Replica Set to a Replicated Sharded Cluster (page 511) Add Shards to a Cluster (page 510) Remove Shards from an Existing Sharded Cluster (page 530) Deploy Three Cong Servers for Production Deployments (page 518) Migrate Cong Servers with the Same Hostname (page 519) Migrate Cong Servers with Different Hostnames (page 519) Replace a Cong Server (page 520) Backup Cluster Metadata (page 521) Backup a Small Sharded Cluster with mongodump (page 54) Create Backup of a Sharded Cluster with Filesystem Snapshots (page 55) Create Backup of a Sharded Cluster with Database Dumps (page 56) Restore a Single Shard (page 58) Restore Sharded Clusters (page 58) Schedule Backup Window for Sharded Clusters (page 59) Administer and Manage Shard Tags (page 517)
17.2.3 Basic Operations

Use Database Commands (page 121) Recover MongoDB Data following Unexpected Shutdown (page 52) Copy Databases Between Instances (page 50) 124 Chapter 17. MongoDB Tutorials
Expire Data from Collections by Setting TTL (page 577) Analyze Performance of Database Operations (page 99) Rotate Log Files (page 89) Build Old Style Indexes (page 356) Manage mongod Processes (page 85) Use mongodump and mongorestore to Backup and Restore MongoDB Databases (page 43) Use Filesystem Snapshots to Backup and Restore MongoDB Databases (page 46)
17.2.4 Security
Congure Linux iptables Firewall for MongoDB (page 137) Congure Windows netsh Firewall for MongoDB (page 141) Enable Authentication (page 146) Create an Administrator with Superuser Credentials (page 146) Add a User to a Database (page 148) Generate a Key File (page 148) Deploy MongoDB with Kerberos Authentication (page 149) Create a Vulnerability Report (page 144)
17.3 Development Patterns

Perform Two Phase Commits (page 563) Isolate Sequence of Operations (page 571) Create an Auto-Incrementing Sequence Field (page 572) Enforce Unique Keys for Sharded Collections (page 533) Aggregation Framework Examples (page 261) Model Data to Support Keyword Search (page 247) Limit Number of Elements in an Array after an Update (page 575) Perform Incremental Map-Reduce (page 313) Troubleshoot the Map Function (page 317) Troubleshoot the Reduce Function (page 318) Store a JavaScript Function on the Server (page 561)
17.4 Application Development

Write a Tumblelog Application with Django MongoDB Engine (page 677) Write a Tumblelog Application with Flask and MongoEngine (page 689)
17.3. Development Patterns
125
17.5 Text Search Patterns

Enable Text Search (page 371) Search String Content for Text (page 378) Create text Index on Multiple Fields (page 372) Specify a Language for Text Index (page 373) Specify text Index Name to Avoid Name Length Limit (page 373) Control Results of Text Search with Weights (page 375) Create a text Index on a Multi-language Collection (page 374) Return Text Queries Using Only a text Index (page 378) Limit the Number of Index Entries Scanned for Text Search (page 376)
17.6 Data Modeling Patterns

Model Embedded One-to-One Relationships Between Documents (page 239) Model Embedded One-to-Many Relationships Between Documents (page 240) Model Referenced One-to-Many Relationships Between Documents (page 241) Model Data for Atomic Operations (page 243) Model Tree Structures with Parent References (page 244) Model Tree Structures with Child References (page 244) Model Tree Structures with Materialized Paths (page 246) Model Tree Structures with Nested Sets (page 247)
17.7 MongoDB Use Case Studies

Storing Log Data (page 609) Pre-Aggregated Reports (page 619) Hierarchical Aggregation (page 628) Product Catalog (page 637) Inventory Management (page 645) Category Hierarchy (page 651) Metadata and Asset Management (page 659) Storing Comments (page 666) See also: Sharded Cluster Administration (page 503) Replica Set Administration (page 419) Indexing Operations (page 347)
126
Chapter 17. MongoDB Tutorials
Part III
Security
127
The documentation in this section outlines basic security, risk management, and access control, and includes specic tasks for conguring rewalls, authentication, and system privileges. User roles in MongoDB provide granular control over user authorization and access. If you believe you have discovered a vulnerability in MongoDB, please see Create a Vulnerability Report (page 144).
129
130
CHAPTER 18
Security Concepts and Strategies
18.1 Security Practices and Management

This document describes risk mitigation in MongoDB deployments. As with all software running in a networked environment, administrators of MongoDB must consider security and risk exposures for a MongoDB deployment. There are no magic solutions for risk mitigation, and maintaining a secure MongoDB deployment is an ongoing process. This document takes a Defense in Depth approach to securing MongoDB deployments and addresses a number of different methods for managing risk and reducing risk exposure. The intent of a Defense In Depth approach is to ensure there are no exploitable points of failure in your deployment that could allow an intruder or un-trusted party to access the data stored in the MongoDB database. The easiest and most effective way to reduce the risk of exploitation is to run MongoDB in a trusted environment, limit access, follow a system of least privilege, and follow best development and deployment practices. See the Strategies for Reducing Risk (page 131) section. For an outline of all security, authentication, and authorization documentation, see Security (page 129).
18.1.1 Strategies for Reducing Risk

The most effective way to reduce risk for MongoDB deployments is to run your entire MongoDB deployment, including all MongoDB components (i.e. mongod (page 1021), mongos (page 1032) and application instances) in a trusted environment. Trusted environments use the following strategies to control access: network lter (e.g. rewall) rules that block all connections from unknown systems to MongoDB components. bind mongod (page 1021) and mongos (page 1032) instances to specic IP addresses to limit accessibility. limit MongoDB programs to non-public local networks, and virtual private networks. You may further reduce risk by: requiring authentication (page 135) for access to MongoDB instances. requiring strong, complex, single purpose authentication credentials. This should be part of your internal security policy. deploying a model of least privilege, where all users have only the amount of access they need to accomplish required tasks, and no more. following the best application development and deployment practices, which includes: validating all inputs, managing sessions, and application-level access control. Continue reading this document for more information on specic strategies and congurations to help reduce the risk exposure of your application.
131
18.1.2 Vulnerability Notication

10gen takes the security of MongoDB and associated products very seriously. If you discover a vulnerability in MongoDB or another 10gen product, or would like to know more about our vulnerability reporting and response process, see the Create a Vulnerability Report (page 144) document.
18.1.3 Runtime Security Conguration

For conguration settings that affect security, see Security Considerations (page 36).
18.1.4 Networking Risk Exposure

Interfaces and Port Numbers The following list includes all default ports used by MongoDB: 27017 This is the default port for mongod (page 1021) and mongos (page 1032) instances. You can change this port with port (page 1079) or --port (page 1022). 27018 This is the default port when running with --shardsvr (page 1030) runtime operation or shardsvr (page 1089) setting. 27019 This is the default port when running with --configsvr (page 1030) runtime operation or configsvr (page 1089) setting. 28017 This is the default port for the web status page. This is always accessible at a port that is 1000 greater than the port determined by port (page 1079). By default MongoDB programs (i.e. mongos (page 1032) and mongod (page 1021)) will bind to all available network interfaces (i.e. IP addresses) on a system. The next section outlines various runtime options that allow you to limit access to MongoDB programs. Network Interface Limitation You can limit the network exposure with the following conguration options: the nohttpinterface (page 1084) setting for mongod (page 1021) and mongos (page 1032) instances. Disables the home status page, which would run on port 28017 by default. The status interface is readonly by default. You may also specify this option on the command line as mongod --nohttpinterface (page 1026) or mongos --nohttpinterface (page 1035). Authentication does not control or affect access to this interface. Important: Disable this option for production deployments. If you do leave this interface enabled, you should only allow trusted clients to access this port. See Firewalls (page 133). the port (page 1079) setting for mongod (page 1021) and mongos (page 1032) instances. Changes the main port on which the mongod (page 1021) or mongos (page 1032) instance listens for connections. Changing the port does not meaningfully reduce risk or limit exposure. You may also specify this option on the command line as mongod --port (page 1022) or mongos --port (page 1032). Whatever port you attach mongod (page 1021) and mongos (page 1032) instances to, you should only allow trusted clients to connect to this port. See Firewalls (page 133).
132
Chapter 18. Security Concepts and Strategies
the rest (page 1085) setting for mongod (page 1021). Enables a fully interactive administrative REST interface, which is disabled by default. The status interface, which is enabled by default, is read-only. This conguration makes that interface fully interactive. The REST interface does not support any authentication and you should always restrict access to this interface to only allow trusted clients to connect to this port. You may also enable this interface on the command line as mongod --rest (page 1026). Important: Disable this option for production deployments. If do you leave this interface enabled, you should only allow trusted clients to access this port. the bind_ip (page 1079) setting for mongod (page 1021) and mongos (page 1032) instances. Limits the network interfaces on which MongoDB programs will listen for incoming connections. You can also specify a number of interfaces by passing bind_ip (page 1079) a comma separated list of IP addresses. You can use the mongod --bind_ip (page 1022) and mongos --bind_ip (page 1032) option on the command line at run time to limit the network accessibility of a MongoDB program. Important: Make sure that your mongod (page 1021) and mongos (page 1032) instances are only accessible on trusted networks. If your system has more than one network interface, bind MongoDB programs to the private or internal network interface.
Firewalls Firewalls allow administrators to lter and control access to a system by providing granular control over what network communications. For administrators of MongoDB, the following capabilities are important: limiting incoming trafc on a specic port to specic systems. limiting incoming trafc from untrusted hosts. On Linux systems, the iptables interface provides access to the underlying netfilter rewall. On Windows systems netsh command line interface provides access to the underlying Windows Firewall. For additional information about rewall conguration consider the following documents: Congure Linux iptables Firewall for MongoDB (page 137) Congure Windows netsh Firewall for MongoDB (page 141) For best results and to minimize overall exposure, ensure that only trafc from trusted sources can reach mongod (page 1021) and mongos (page 1032) instances and that the mongod (page 1021) and mongos (page 1032) instances can only connect to trusted outputs. See also: For MongoDB deployments on Amazons web services, see the Amazon EC2 page, which addresses Amazons Security Groups and other EC2-specic security features. Virtual Private Networks Virtual private networks, or VPNs, make it possible to link two networks over an encrypted and limited-access trusted network. Typically MongoDB users who use VPNs use SSL rather than IPSEC VPNs for performance issues. Depending on conguration and implementation VPNs provide for certicate validation and a choice of encryption protocols, which requires a rigorous level of authentication and identication of all clients. Furthermore, because VPNs provide a secure tunnel, using a VPN connection to control access to your MongoDB instance, you can prevent tampering and man-in-the-middle attacks. 18.1. Security Practices and Management 133
18.1.5 Operations
Always run the mongod (page 1021) or mongos (page 1032) process as a unique user with the minimum required permissions and access. Never run a MongoDB program as a root or administrative users. The system users that run the MongoDB processes should have robust authentication credentials that prevent unauthorized or casual access. To further limit the environment, you can run the mongod (page 1021) or mongos (page 1032) process in a chroot environment. Both user-based access restrictions and chroot conguration follow recommended conventions for administering all daemon processes on Unix-like systems. You can disable anonymous access to the database by enabling MongoDB authentication. See Access Control (page 135).
18.1.6 Interfaces
Simply limiting access to a mongod (page 1021) is not sufcient for totally controlling risk exposure. Consider the recommendations in the following section, for limiting exposure other interface-related risks. JavaScript and the Security of the mongo Shell Be aware of the following capabilities and behaviors of the mongo (page 1036) shell: mongo (page 1036) will evaluate a .js le passed to the mongo --eval (page 1037) option. The mongo (page 1036) shell does not validate the input of JavaScript input to --eval (page 1037). mongo (page 1036) will evaluate a .mongorc.js le before starting. You can disable this behavior by passing the mongo --norc (page 1036) option. On Linux and Unix systems, mongo (page 1036) reads the .mongorc.js le from $HOME /.mongorc.js (i.e. ~/.mongorc.js), and Windows mongo.exe reads the .mongorc.js le from %HOME%.mongorc.js or %HOMEDRIVE%%HOMEPATH%.mongorc.js. HTTP Status Interface The HTTP status interface provides a web-based interface that includes a variety of operational data, logs, and status reports regarding the mongod (page 1021) or mongos (page 1032) instance. The HTTP interface is always available on the port numbered 1000 greater than the primary mongod (page 1021) port. By default this is 28017, but is indirectly set using the port (page 1079) option which allows you to congure the primary mongod (page 1021) port. Without the rest (page 1085) setting, this interface is entirely read-only, and limited in scope; nevertheless, this interface may represent an exposure. To disable the HTTP interface, set the nohttpinterface (page 1084) run time option or the --nohttpinterface (page 1026) command line option. REST API The REST API to MongoDB provides additional information and write access on top of the HTTP Status interface. The REST interface is disabled by default, and is not recommended for production use. While the REST API does not provide any support for insert, update, or remove operations, it does provide administrative access, and its accessibility represents a vulnerability in a secure environment. If you must use the REST API, please control and limit access to the REST API. The REST API does not include any support for authentication, even when running with auth (page 1081) enabled. See the following documents for instructions on restricting access to the REST API interface: 134 Chapter 18. Security Concepts and Strategies
Congure Linux iptables Firewall for MongoDB (page 137) Congure Windows netsh Firewall for MongoDB (page 141)
18.1.7 Data Encryption

To support audit requirements, you may need to encrypt data stored in MongoDB. For best results you can encrypt this data in the application layer, by encrypting the content of elds that hold secure data. Additionally, 10gen has a partnership with Gazzang to encrypt and secure sensitive data within MongoDB. The solution encrypts data in real time and Gazzang provides advanced key management that ensures only authorized processes and can access this data. The Gazzang software ensures that the cryptographic keys remain safe and ensures compliance with standards including HIPAA, PCI-DSS, and FERPA. For more information consider the following resources: Datasheet Webinar
18.2 Access Control

MongoDB provides support for authentication and authorization by storing a users credentials and privileges in a databases system.users (page 160) collection. MongoDB provisions authentication and access on a per-database level. Users exist in the context of a single logical database. For MongoDB Enterprise installations, MongoDB also provides support for authentication using a Kerberos service. See Deploy MongoDB with Kerberos Authentication (page 149).
18.2.1 Authentication
MongoDB provides support for basic authentication by: storing user credentials in a databases system.users (page 160) collection, and providing the auth (page 1081) and keyFile (page 1081) conguration settings to enable authentication for a given mongod (page 1021) or mongos (page 1032) instance. Authentication is disabled by default. To enable authentication, see the following: Enable Authentication (page 146) Deploy MongoDB with Kerberos Authentication (page 149)
18.2.2 Authorization
MongoDB supports role-based access to databases and database operations by storing each users roles in a privilege document (page 155) in the system.users (page 160) collection. For a description of privilege documents and of available roles, see User Privilege Roles in MongoDB (page 155). Changed in version 2.4: The schema of system.users (page 160) changed to accommodate a more sophisticated user privilege model, as dened in privilege documents (page 155). The system.users (page 160) collection is protected to prevent privilege escalation attacks. To access the collection, you must have the userAdmin (page 157) or userAdminAnyDatabase (page 159) role.
18.2. Access Control
135
To assign user roles, you must rst create an admin user in the database. Then you create additional users, assigning them appropriate user roles. To assign user roles, see the following: Create an Administrator with Superuser Credentials (page 146) Add a User to a Database (page 148) User Roles in the admin Database The admin database provides roles not available in other databases, including a role that effectively makes a user a MongoDB system superuser. See Database Administration Roles (page 156) and Administrative Roles (page 157). Authentication to One Database at a Time You can log in as only one user for a given database, including the admin database. If you authenticate to a database as one user and later authenticate on the same database as a different user, the second authentication invalidates the rst. Logging into a different database, however, does not invalidate authentication on other databases.
136
Chapter 18. Security Concepts and Strategies
CHAPTER 19
Tutorials
19.1 Network Security

19.1.1 Congure Linux iptables Firewall for MongoDB
On contemporary Linux systems, the iptables program provides methods for managing the Linux Kernels netfilter or network packet ltering capabilities. These rewall rules make it possible for administrators to control what hosts can connect to the system, and limit risk exposure by limiting the hosts that can connect to a system. This document outlines basic rewall congurations for iptables rewalls on Linux. Use these approaches as a starting point for your larger networking organization. For a detailed over view of security practices and risk management for MongoDB, see Security Practices and Management (page 131). See also: For MongoDB deployments on Amazons web services, see the Amazon EC2 page, which addresses Amazons Security Groups and other EC2-specic security features. Overview Rules in iptables congurations fall into chains, which describe the process for ltering and processing specic streams of trafc. Chains have an order, and packets must pass through earlier rules in a chain to reach later rules. This document only the following two chains: INPUT Controls all incoming trafc. OUTPUT Controls all outgoing trafc. Given the default ports (page 132) of all MongoDB processes, you must congure networking rules that permit only required communication between your application and the appropriate mongod (page 1021) and mongos (page 1032) instances. Be aware that, by default, the default policy of iptables is to allow all connections and trafc unless explicitly disabled. The conguration changes outlined in this document will create rules that explicitly allow trafc from specic addresses and on specic ports, using a default policy that drops all trafc that is not explicitly allowed. When you have properly congured your iptables rules to allow only the trafc that you want to permit, you can Change Default Policy to DROP (page 140).
137
Patterns This section contains a number of patterns and examples for conguring iptables for use with MongoDB deployments. If you have congured different ports using the port (page 1079) conguration setting, you will need to modify the rules accordingly.
Trafc to and from mongod Instances
This pattern is applicable to all mongod (page 1021) instances running as standalone instances or as part of a replica set. The goal of this pattern is to explicitly allow trafc to the mongod (page 1021) instance from the application server. In the following examples, replace <ip-address> with the IP address of the application server:
iptables -A INPUT -s <ip-address> -p tcp --destination-port 27017 -m state --state NEW,ESTABLISHED -j iptables -A OUTPUT -d <ip-address> -p tcp --source-port 27017 -m state --state ESTABLISHED -j ACCEPT
The rst rule allows all incoming trafc from <ip-address> on port 27017, which allows the application server to connect to the mongod (page 1021) instance. The second rule, allows outgoing trafc from the mongod (page 1021) to reach the application server. Optional If you have only one application server, you can replace <ip-address> with either the IP address itself, such as: 198.51.100.55. You can also express this using CIDR notation as 198.51.100.55/32. If you want to permit a larger block of possible IP addresses you can allow trafc from a http://docs.mongodb.org/manual/24 using one of the following specications for the <ip-address>, as follows:
10.10.10.10/24 10.10.10.10/255.255.255.0
Trafc to and from mongos Instances
mongos (page 1032) instances provide query routing for sharded clusters. Clients connect to mongos (page 1032) instances, which behave from the clients perspective as mongod (page 1021) instances. In turn, the mongos (page 1032) connects to all mongod (page 1021) instances that are components of the sharded cluster. Use the same iptables command to allow trafc to and from these instances as you would from the mongod (page 1021) instances that are members of the replica set. Take the conguration outlined in the Trafc to and from mongod Instances (page 138) section as an example.
Trafc to and from a MongoDB Cong Server
Cong servers, host the cong database that stores metadata for sharded clusters. Each production cluster has three cong servers, initiated using the mongod --configsvr (page 1030) option. 1 Cong servers listen for connections on port 27019. As a result, add the following iptables rules to the cong server to allow incoming and outgoing connection on port 27019, for connection to the other cong servers.
1
You can also run a cong server by setting the configsvr (page 1089) option in a conguration le.
138
Chapter 19. Tutorials
Replace <ip-address> with the address or address space of all the mongod (page 1021) that provide cong servers. Additionally, cong servers need to allow incoming connections from all of the mongos (page 1032) instances in the cluster and all mongod (page 1021) instances in the cluster. Add rules that resemble the following:
iptables -A INPUT -s <ip-address> -p tcp --destination-port 27019 -m state --state NEW,ESTABLISHED -j
Replace <ip-address> with the address of the mongos (page 1032) instances and the shard mongod (page 1021) instances.
Trafc to and from a MongoDB Shard Server
For shard servers, running as mongod --shardsvr (page 1030) 2 Because the default port number when running with shardsvr (page 1089) is 27018, you must congure the following iptables rules to allow trafc to and from each shard:
Replace the <ip-address> specication with the IP address of all mongod (page 1021). This allows you to permit incoming and outgoing trafc between all shards including constituent replica set members, to: all mongod (page 1021) instances in the shards replica sets. all mongod (page 1021) instances in other shards.
3
Furthermore, shards need to be able make outgoing connections to: all mongos (page 1032) instances. all mongod (page 1021) instances in the cong servers. Create a rule that resembles the following, and replace the <ip-address> with the address of the cong servers and the mongos (page 1032) instances:
iptables -A OUTPUT -d <ip-address> -p tcp --source-port 27018 -m state --state ESTABLISHED -j ACCEPT
Provide Access For Monitoring Systems
1. The mongostat (page 1063) diagnostic tool, when running with the --discover (page 1065) needs to be able to reach all components of a cluster, including the cong servers, the shard servers, and the mongos (page 1032) instances. 2. If your monitoring system needs access the HTTP interface, insert the following rule to the chain:
iptables -A INPUT -s <ip-address> -p tcp --destination-port 28017 -m state --state NEW,ESTABLISH
Replace <ip-address> with the address of the instance that needs access to the HTTP or REST interface. For all deployments, you should restrict access to this port to only the monitoring instance. Optional For shard server mongod (page 1021) instances running with shardsvr (page 1089), the rule would resemble the following:
2 You can also specify the shard server option using the shardsvr (page 1089) setting in the conguration le. Shard members are also often conventional replica sets using the default port. 3 All shards in a cluster need to be able to communicate with all other shards to facilitate chunk and balancing operations.
19.1. Network Security
139
For cong server mongod (page 1021) instances running with configsvr (page 1089), the rule would resemble the following:
Change Default Policy to DROP The default policy for iptables chains is to allow all trafc. After completing all iptables conguration changes, you must change the default policy to DROP so that all trafc that isnt explicitly allowed as above will not be able to reach components of the MongoDB deployment. Issue the following commands to change this policy:
iptables -P INPUT DROP iptables -P OUTPUT DROP
Manage and Maintain iptables Conguration This section contains a number of basic operations for managing and using iptables. There are various front end tools that automate some aspects of iptables conguration, but at the core all iptables front ends provide the same basic functionality:
Make all iptables Rules Persistent
By default all iptables rules are only stored in memory. When your system restarts, your rewall rules will revert to their defaults. When you have tested a rule set and have guaranteed that it effectively controls trafc you can use the following operations to you should make the rule set persistent. On Red Hat Enterprise Linux, Fedora Linux, and related distributions you can issue the following command:
service iptables save
On Debian, Ubuntu, and related distributions, you can use the following command to dump the iptables rules to the http://docs.mongodb.org/manual/etc/iptables.conf le:
iptables-save > /etc/iptables.conf
Run the following operation to restore the network rules:

iptables-restore < /etc/iptables.conf
Place this command in your rc.local le, or in the http://docs.mongodb.org/manual/etc/network/if-up.d/ipta le with other similar operations.q
List all iptables Rules
To list all of currently applied iptables rules, use the following operation at the system shell.
iptables --L
140
Flush all iptables Rules
If you make a conguration mistake when entering iptables rules or simply need to revert to the default rule set, you can use the following operation at the system shell to ush all rules:
iptables --F
If youve already made your iptables rules persistent, you will need to repeat the appropriate procedure in the Make all iptables Rules Persistent (page 140) section.
19.1.2 Congure Windows netsh Firewall for MongoDB

On Windows Server systems, the netsh program provides methods for managing the Windows Firewall. These rewall rules make it possible for administrators to control what hosts can connect to the system, and limit risk exposure by limiting the hosts that can connect to a system. This document outlines basic Windows Firewall congurations. Use these approaches as a starting point for your larger networking organization. For a detailed over view of security practices and risk management for MongoDB, see Security Practices and Management (page 131). See also: Windows Firewall documentation from Microsoft. Overview Windows Firewall processes rules in an ordered determined by rule type, and parsed in the following order: 1. Windows Service Hardening 2. Connection security rules 3. Authenticated Bypass Rules 4. Block Rules 5. Allow Rules 6. Default Rules By default, the policy in Windows Firewall allows all outbound connections and blocks all incoming connections. Given the default ports (page 132) of all MongoDB processes, you must congure networking rules that permit only required communication between your application and the appropriate mongod.exe (page 1041) and mongos.exe (page 1042) instances. The conguration changes outlined in this document will create rules which explicitly allow trafc from specic addresses and on specic ports, using a default policy that drops all trafc that is not explicitly allowed. You can congure the Windows Firewall with using the netsh command line tool or through a windows application. On Windows Server 2008 this application is Windows Firewall With Advanced Security in Administrative Tools. On previous versions of Windows Server, access the Windows Firewall application in the System and Security control panel. The procedures in this document use the netsh command line tool.
141
Patterns This section contains a number of patterns and examples for conguring Windows Firewall for use with MongoDB deployments. If you have congured different ports using the port (page 1079) conguration setting, you will need to modify the rules accordingly.
Trafc to and from mongod.exe Instances
This pattern is applicable to all mongod.exe (page 1041) instances running as standalone instances or as part of a replica set. The goal of this pattern is to explicitly allow trafc to the mongod.exe (page 1041) instance from the application server.
netsh advfirewall firewall add rule name="Open mongod port 27017" dir=in action=allow protocol=TCP lo
This rule allows all incoming trafc to port 27017, which allows the application server to connect to the mongod.exe (page 1041) instance. Windows Firewall also allows enabling network access for an entire application rather than to a specic port, as in the following example:
netsh advfirewall firewall add rule name="Allowing mongod" dir=in action=allow program=" C:\mongodb\b
You can allow all access for a mongos.exe (page 1042) server, with the following invocation:
netsh advfirewall firewall add rule name="Allowing mongos" dir=in action=allow program=" C:\mongodb\b
Trafc to and from mongos.exe Instances
mongos.exe (page 1042) instances provide query routing for sharded clusters. Clients connect to mongos.exe (page 1042) instances, which behave from the clients perspective as mongod.exe (page 1041) instances. In turn, the mongos.exe (page 1042) connects to all mongod.exe (page 1041) instances that are components of the sharded cluster. Use the same Windows Firewall command to allow trafc to and from these instances as you would from the mongod.exe (page 1041) instances that are members of the replica set.
netsh advfirewall firewall add rule name="Open mongod shard port 27018" dir=in action=allow protocol=
Trafc to and from a MongoDB Cong Server
Conguration servers, host the cong database that stores metadata for sharded clusters. Each production cluster has three conguration servers, initiated using the mongod --configsvr (page 1030) option. 4 Conguration servers listen for connections on port 27019. As a result, add the following Windows Firewall rules to the cong server to allow incoming and outgoing connection on port 27019, for connection to the other cong servers.
netsh advfirewall firewall add rule name="Open mongod config svr port 27019" dir=in action=allow prot
Additionally, cong servers need to allow incoming connections from all of the mongos.exe (page 1042) instances in the cluster and all mongod.exe (page 1041) instances in the cluster. Add rules that resemble the following:
netsh advfirewall firewall add rule name="Open mongod config svr inbound" dir=in action=allow protoco
Replace <ip-address> with the addresses of the mongos.exe (page 1042) instances and the shard mongod.exe (page 1041) instances.
4
You can also run a cong server by setting the configsvr (page 1089) option in a conguration le.
142
Trafc to and from a MongoDB Shard Server
For shard servers, running as mongod --shardsvr (page 1030) 5 Because the default port number when running with shardsvr (page 1089) is 27018, you must congure the following Windows Firewall rules to allow trafc to and from each shard:
netsh advfirewall firewall add rule name="Open mongod shardsvr inbound" dir=in action=allow protocol= netsh advfirewall firewall add rule name="Open mongod shardsvr outbound" dir=out action=allow protoco
Replace the <ip-address> specication with the IP address of all mongod.exe (page 1041) instances. This allows you to permit incoming and outgoing trafc between all shards including constituent replica set members to: all mongod.exe (page 1041) instances in the shards replica sets. all mongod.exe (page 1041) instances in other shards.
6
Furthermore, shards need to be able make outgoing connections to: all mongos.exe (page 1042) instances. all mongod.exe (page 1041) instances in the cong servers. Create a rule that resembles the following, and replace the <ip-address> with the address of the cong servers and the mongos.exe (page 1042) instances:
netsh advfirewall firewall add rule name="Open mongod config svr outbound" dir=out action=allow proto
Provide Access For Monitoring Systems
1. The mongostat (page 1063) diagnostic tool, when running with the --discover (page 1065) needs to be able to reach all components of a cluster, including the cong servers, the shard servers, and the mongos.exe (page 1042) instances. 2. If your monitoring system needs access the HTTP interface, insert the following rule to the chain:
netsh advfirewall firewall add rule name="Open mongod HTTP monitoring inbound" dir=in action=all
Replace <ip-address> with the address of the instance that needs access to the HTTP or REST interface. For all deployments, you should restrict access to this port to only the monitoring instance. Optional For shard server mongod.exe (page 1041) instances running with shardsvr (page 1089), the rule would resemble the following:
netsh advfirewall firewall add rule name="Open mongos HTTP monitoring inbound" dir=in action=all
For cong server mongod.exe (page 1041) instances running with configsvr (page 1089), the rule would resemble the following:
netsh advfirewall firewall add rule name="Open mongod configsvr HTTP monitoring inbound" dir=in
5 You can also specify the shard server option using the shardsvr (page 1089) setting in the conguration le. Shard members are also often conventional replica sets using the default port. 6 All shards in a cluster need to be able to communicate with all other shards to facilitate chunk and balancing operations.
143
Manage and Maintain Windows Firewall Congurations This section contains a number of basic operations for managing and using netsh. While you can use the GUI front ends to manage the Windows Firewall, all core functionality is accessible is accessible from netsh.
Delete all Windows Firewall Rules
To delete the rewall rule allowing mongod.exe (page 1041) trafc:

netsh advfirewall firewall delete rule name="Open mongod port 27017" protocol=tcp localport=27017
netsh advfirewall firewall delete rule name="Open mongod shard port 27018" protocol=tcp localport=270
List All Windows Firewall Rules
To return a list of all Windows Firewall rules:

netsh advfirewall firewall show rule name=all
Reset Windows Firewall
To reset the Windows Firewall rules:

netsh advfirewall reset
Backup and Restore Windows Firewall Rules
To simplify administration of larger collection of systems, you can export or import rewall systems from different servers) rules very easily on Windows: Export all rewall rules with the following command:
netsh advfirewall export "C:\temp\MongoDBfw.wfw"
Replace "C:\temp\MongoDBfw.wfw" with a path of your choosing. You can use a command in the following form to import a le created using this operation:
netsh advfirewall import "C:\temp\MongoDBfw.wfw"
19.1.3 Create a Vulnerability Report

If you believe you have discovered a vulnerability in MongoDB or a related product or have experienced a security incident related to MongoDB, please report the issue so that 10gen can respond appropriately and work to prevent additional issues in the future. To report an issue, use either jira.mongodb.org (preferred) or email. 10gen responds to vulnerability notications within 48 hours.
144
Information to Provide All vulnerability reports should contain as much information as possible so 10gen can move quickly to resolve the issue. In particular, please include the following: The name of the product. Common Vulnerability information, if applicable, including: CVSS (Common Vulnerability Scoring System) Score. CVE (Common Vulnerability and Exposures) Identier. Contact information, including an email address and/or phone number, if applicable. Create the Report in Jira 10gen prefers jira.mongodb.org for all communication regarding MongoDB and related products. Submit a ticket in the Core Server Security project at: https://jira.mongodb.org/browse/SECURITY/. The ticket number will become the reference identication for the issue for the lifetime of the issue. You can use this identier for tracking purposes. Send the Report via Email While Jira is preferred, you may also report vulnerabilities via email to security@10gen.com. You may encrypt email using the 10gen public key at http://docs.mongodb.org/10gen-gpg-key.asc. 10gen responds to vulnerability reports sent via email with a response email that contains a reference number for a Jira ticket posted to the SECURITY project. Evaluation of a Vulnerability Report 10gen validates all submitted vulnerabilities and uses Jira to track all communications regarding a vulnerability, including requests for clarication or additional information. If needed, 10gen representatives set up a conference call to exchange information regarding the vulnerability. Disclosure 10gen requests that you do not publicly disclose any information regarding the vulnerability or exploit the issue until 10gen has had the opportunity to analyze the vulnerability, to respond to the notication, and to notify key users, customers, and partners. The amount of time required to validate a reported vulnerability depends on the complexity and severity of the issue. 10gen takes all required vulnerabilities very seriously and will always ensure that there is a clear and open channel of communication with the reporter. After validating an issue, 10gen coordinates public disclosure of the issue with the reporter in a mutually agreed timeframe and format. If required or requested, the reporter of a vulnerability will receive credit in the published security bulletin.
145
19.2 Access Control

19.2.1 Enable Authentication
Enable authentication using the auth (page 1081) or keyFile (page 1081) settings. Use auth (page 1081) for standalone instances, and keyFile (page 1081) with with replica sets and sharded clusters. keyFile (page 1081) implies auth (page 1081) and allows members of a MongoDB deployment to authenticate internally. Authentication requires at least one administrator user in the admin database. You can create the user before enabling authentication or after enabling authentication. See also: Deploy MongoDB with Kerberos Authentication (page 149). Also consider the password hashing (page 162) issue resolved after 2.2. Procedures You can enable authentication using either of the following procedures, depending
Create the Administrator Credentials and then Enable Authentication
1. Start the mongod (page 1021) or mongos (page 1032) instance without the auth (page 1081) or keyFile (page 1081) setting. 2. Create the administrator user as described in Create an Administrator with Superuser Credentials (page 146). 3. Re-start the mongod (page 1021) or mongos (page 1032) instance with the auth (page 1081) or keyFile (page 1081) setting.
Enable Authentication and then Create Administrator
1. Start the mongod (page 1021) or mongos (page 1032) instance with the auth (page 1081) or keyFile (page 1081) setting. 2. Connect to the instance on the same system so that you can authenticate using the localhost exception (page 147). 3. Create the administrator user as described in Create an Administrator with Superuser Credentials (page 146). Query Authenticated Users If you have the userAdmin (page 157) or userAdminAnyDatabase (page 159) role on a database, you can query authenticated users in that database with the following operation:
db.system.users.find()
19.2.2 Create an Administrator with Superuser Credentials

A MongoDB superuser is a user with either the userAdmin (page 157) or userAdminAnyDatabase (page 159) role. The user with either of these roles can create and modify any other users and can assign them any privileges. The user also can grant itself any privileges. In production deployments, this user should have no other roles and should only administer users and privileges.
146
This should be the rst user created for a MongoDB deployment. This user can then create all other users in the system. Important: The userAdminAnyDatabase (page 159) user can grant itself and any other user full access to anything. The credentials to log in as this user should be carefully controlled. Important: A user with the userAdminAnyDatabase (page 159) is not the same as the traditional root superuser in that this role does not enable the user to read/insert data in the database or perform database-related operations.
Create a UserAdmin Superuser 1. Connect to the mongod (page 1021) or mongos (page 1032) by either: Authenticating as an existing user with the userAdmin (page 157) or userAdminAnyDatabase (page 159) role. Authenticating using the localhost exception (page 147). When creating the rst user in a deployment, you must authenticate using the localhost exception (page 147). 2. Switch to the admin database:
db = db.getSiblingDB(admin)
3. Add the user with either the userAdmin (page 157) role or userAdminAnyDatabase (page 159) role, and only that role, by issuing a command similar to the following, where <username> is the username and <password> is the password:
db.addUser( { user: "<username>", pwd: "<password>", roles: [ "userAdminAnyDatabase" ] } )
Authenticate with Full Administrative Access via Localhost If there are no users for the admin database, you can connect with full administrative access via the localhost interface. This bypass exists to support bootstrapping new deployments. This approach is useful, for example, if you want to run mongod (page 1021) or mongos (page 1032) with authentication before creating your rst user. To authenticate via localhost, connect to the mongod (page 1021) or mongos (page 1032) from a client running on the same system. Your connection will have full administrative access. To disable the localhost bypass, set the enableLocalhostAuthBypass (page 1093) parameter using setParameter (page 1087) during startup:
mongod --setParameter enableLocalhostAuthBypass=0
Note: For versions of MongoDB 2.2 prior to 2.2.4, if mongos (page 1032) is running with keyFile (page 1081), then all users connecting over the localhost interface must authenticate, even if there arent any users in the admin database. Connections on localhost are not correctly granted full access on sharded systems that run those versions. MongoDB 2.2.4 resolves this issue. Note: In version 2.2, you cannot add the rst user to a sharded cluster using the localhost connection. If you are running a 2.2 sharded cluster and want to enable authentication, you must deploy the cluster and add the rst user to the admin database before restarting the cluster to run with keyFile (page 1081).
147
19.2.3 Add a User to a Database

To add a user to a database you must authenticate to that database as a user with the userAdmin (page 157) or userAdminAnyDatabase (page 159) role. If you have not rst created a user with one of those roles, do so as described in Create an Administrator with Superuser Credentials (page 146). When adding a user to multiple databases, you must give the user a unique username and password combination for each database. See Password Hashing Insecurity (page 162) for important security information. To add a user, pass the db.addUser() (page 967) method a well formed privilege document (page 155) that contains the users credentials and privileges. The db.addUser() (page 967) method adds the document to the databases system.users (page 160) collection. For the structure of a privilege document, see system.users (page 160). For descriptions of user roles, see User Privilege Roles in MongoDB (page 155). Example The following creates a user named Alice in the products database and gives her readWrite and dbAdmin privileges.
use products db.addUser( { user: "Alice", pwd: "Moon1234", roles: [ "readWrite", "dbAdmin" ] } )
Example The following creates a user named Bob in the admin database. The privilege document (page 160) uses Bobs credentials from the products database and assigns him userAdmin privileges.
use admin db.addUser( { user: "Bob", userSource: "products", roles: [ "userAdmin" ] } )
Example The following creates a user named Carlos in the admin database and gives him readWrite access to the config database, which lets him change certain settings for sharded clusters, such as to disable the balancer.
db = db.getSiblingDB(admin) db.addUser( { user: "Carlos", pwd: "Moon1234", roles: [ "clusterAdmin" ], otherDBRoles: { config: [ "readWrite" ] } } )
Only the admin database supports the otherDBRoles (page 161) eld.
19.2.4 Generate a Key File

This section describes how to generate a key le to store authentication information. After generating a key le, specify the key le using the keyFile (page 1081) option when starting a mongod (page 1021) or mongos (page 1032)
148
instance. A key le must be less than one kilobyte in size and may only contain characters in the base64 set. The key le must not have group or world permissions on UNIX systems. Key le permissions are not checked on Windows systems. Generate a Key File on a Windows System Use the following openssl command at the system shell to generate pseudo-random content for a key le for deployments with Windows components:
openssl rand -base64 741
Generate a Key File on a Linux or Unix System Use the following openssl command at the system shell to generate pseudo-random content for a key le for systems that do not have Windows components (i.e. OS X, Unix, or Linux systems):
Key File Properties Be aware that MongoDB strips whitespace characters (e.g. x0d, x09, and x20,) for cross-platform convenience. As a result, the following operations produce identical keys:
echo echo echo echo -e -e -e -e "my secret key" > key1 "my secret key\n" > key2 "my secret key" > key3 "my\r\nsecret\r\nkey\r\n" > key4
19.2.5 Deploy MongoDB with Kerberos Authentication

New in version 2.4. MongoDB Enterprise supports authentication using a Kerberos service to manage the authentication process. Kerberos is an industry standard authentication protocol for large client/server system. With Kerberos MongoDB and application ecosystems can take advantage of existing authentication infrastructure and processes. Setting up and conguring a Kerberos deployment is beyond the scope of this document. In order to use MongoDB with Kerberos, you must have a properly congured Kerberos deployment and the ability to generate a valid keytab le for each mongod (page 1021) instance in your MongoDB deployment. Note: The following assumes that you have a valid Kerberos keytab le for your realm accessible on your system. The examples below assume that the keytab le is valid and is located at http://docs.mongodb.org/manual/opt/mongodb/mongod.keytab and is only accessible to the user that runs the mongod (page 1021) process.
Process Overview To run MongoDB with Kerberos support, you must: Congure a Kerberos service principal for each mongod (page 1021) and mongos (page 1032) instance in your MongoDB deployment.
149
Generate and distribute keytab les for each MongoDB component (i.e. mongod (page 1021) and mongos (page 1032))in your deployment. Ensure that you only transmit keytab les over secure channels. Optional. Start the mongod (page 1021) instance without auth (page 1081) and create users inside of MongoDB that you can use to bootstrap your deployment. Start mongod (page 1021) and mongos (page 1032) with the KRB5_KTNAME environment variable as well as a number of required run time options. If you did not create Kerberos user accounts, you can use the localhost exception (page 147) to create users at this point until you create the rst user on the admin database. Authenticate clients, including the mongo (page 1036) shell using Kerberos. Operations
Create Users and Privilege Documents
For every user that you want to be able to authenticate using Kerberos, you must create corresponding privilege documents in the system.users (page 160) collection to provision access to users. Consider the following document:
{ user: "application/reporting@EXAMPLE.NET", roles: ["read"], userSource: "$external" }
This grants the Kerberos user principal application/reporting@EXAMPLE.NET read only access to a database. The userSource (page 161) $external reference allows mongod (page 1021) to consult an external source (i.e. Kerberos) to authenticate this user. In the mongo (page 1036) shell you can pass the db.addUser() (page 967) a user privilege document to provision access to users, as in the following operation:
db = db.getSiblingDB("records") db.addUser( { "user": "application/reporting@EXAMPLE.NET", "roles": [ "read" ], "userSource": "$external" } )
These operations grants the Kerberos user application/reporting@EXAMPLE.NET access to the records database. To remove access to a user, use the remove() (page 944) method, as in the following example:
db.system.users.remove( { user: "application/reporting@EXAMPLE.NET" } )
To modify a user document, use update (page 219) operations on documents in the system.users (page 160) collection. See also: system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155).
Start mongod with Kerberos Support
Once you have provisioned privileges to users in the mongod (page 1021), and obtained a valid keytab le, you must start mongod (page 1021) using a command in the following form: 150 Chapter 19. Tutorials
env KRB5_KTNAME=<path to keytab file> <mongod invocation>
For successful operation with mongod (page 1021) use the following run time options in addition to your normal default conguration options: --setParameter (page 1027) with the authenticationMechanisms=GSSAPI argument to enable support for Kerberos. --auth (page 1023) to enable authentication. --keyFile (page 1023) to allow components of a single MongoDB deployment to communicate with each other, if needed to support replica set and sharded cluster operations. keyFile (page 1081) implies auth (page 1081). For example, consider the following invocation:
env KRB5_KTNAME=/opt/mongodb/mongod.keytab \ /opt/mongodb/bin/mongod --dbpath /opt/mongodb/data \ --fork --logpath /opt/mongodb/log/mongod.log \ --auth --setParameter authenticationMechanisms=GSSAPI
You can also specify these options using the conguration le. As in the following:
# /opt/mongodb/mongod.conf, Example configuration file. fork = true auth = true dbpath = /opt/mongodb/data logpath = /opt/mongodb/log/mongod.log setParameter = authenticationMechanisms=GSSAPI
To use this conguration le, start mongod (page 1021) as in the following:
env KRB5_KTNAME=/opt/mongodb/mongod.keytab \ /opt/mongodb/bin/mongod --config /opt/mongodb/mongod.conf
To start a mongos (page 1032) instance using Kerberos, you must create a Kerberos service principal and deploy a keytab le for this instance, and then start the mongos (page 1032) with the following invocation:
env KRB5_KTNAME=/opt/mongodb/mongos.keytab \ /opt/mongodb/bin/mongos --configdb shard0.example.net,shard1.example.net,shard2.example.net \ --setParameter authenticationMechanisms=GSSAPI \ --keyFile /opt/mongodb/mongos.keyfile
If you encounter problems when trying to start mongod (page 1021) or mongos (page 1032), please see the troubleshooting section (page 152) for more information. Important: Before users can authenticate to MongoDB using Kerberos you must create users (page 150) and grant them privileges within MongoDB. If you have not created users when you start MongoDB with Kerberos you can use the localhost authentication exception (page 147) to add users. See the Create Users and Privilege Documents (page 150) section and the User Privilege Roles in MongoDB (page 155) document for more information.
Authenticate mongo Shell with Kerberos
To connect to a mongod (page 1021) instance using the mongo (page 1036) shell you must begin by using the kinit program to initialize and authenticate a Kerberos session. Then, start a mongo (page 1036) instance, and use
151
the db.auth() (page 968) method, to authenticate against the special $external database, as in the following operation:
use $external db.auth( { mechanism: "GSSAPI", user: "application/reporting@EXAMPLE.NET" } )
Alternately, you can authenticate using command line options to mongo (page 1036), as in the following equivalent example:
mongo --authenticationMechanism=GSSAPI --authenticationDatabase=$external \ --username application/reporting@EXAMPLE.NET
These operations authenticates the Kerberos principal name application/reporting@EXAMPLE.NET to the connected mongod (page 1021), and will automatically acquire all available privileges as needed.
Use MongoDB Drivers to Authenticate with Kerberos
At the time of release, the C++, Java, and C# drivers all provide support for Kerberos authentication to MongoDB. Consider the following tutorials for more information: Java C# C++ Troubleshooting
Kerberos Conguration Checklist
If youre having trouble getting mongod (page 1021) to start with Kerberos, there are a number of Kerberos-specic issues that can prevent successful authentication. As you begin troubleshooting your Kerberos deployment, ensure that: The mongod (page 1021) is from MongoDB Enterprise. You have a valid keytab le specied in the environment running the mongod (page 1021). For the mongod (page 1021) instance running on the db0.example.net host, the service principal should be mongodb/db0.example.net. DNS allows the mongod (page 1021) to resolve the components of the Kerberos infrastructure. You should have both A and PTR records (i.e. forward and reverse DNS) for the system that runs the mongod (page 1021) instance. The canonical system hostname of the system that runs the mongod (page 1021) instance is the resolvable fully qualied domain for this host. Test system hostname resolution with the hostname -f command at the system prompt. Both the Kerberos KDC and the system running mongod (page 1021) instance must be able to resolve each other using DNS 7 The time systems of the systems running the mongod (page 1021) instances and the Kerberos infrastructure are synchronized. Time differences greater than 5 minutes will prevent successful authentication.
7 By default, Kerberos attempts to resolve hosts using the content of the http://docs.mongodb.org/manual/etc/kerb5.conf before using DNS to resolve hosts.
152
If you still encounter problems with Kerberos, you can start both mongod (page 1021) and mongo (page 1036) (or another client) with the environment variable KRB5_TRACE set to different les to produce more verbose logging of the Kerberos process to help further troubleshooting, as in the following example:
env KRB5_KTNAME=/opt/mongodb/mongod.keytab \ KRB5_TRACE=/opt/mongodb/log/mongodb-kerberos.log \ /opt/mongodb/bin/mongod --dbpath /opt/mongodb/data \ --fork --logpath /opt/mongodb/log/mongod.log \ --auth --setParameter authenticationMechanisms=GSSAPI
Common Error Messages
In some situations, MongoDB will return error messages from the GSSAPI interface if there is a problem with the Kerberos service. GSSAPI error in client while negotiating security context. This error occurs on the client and reects insufcient credentials or a malicious attempt to authenticate. If you receive this error ensure that youre using the correct credentials and the correct fully qualied domain name when connecting to the host. GSSAPI error acquiring credentials. This error only occurs when attempting to start the mongod (page 1021) or mongos (page 1032) and reects improper conguration of system hostname or a missing or incorrectly congured keytab le. If you encounter this problem, consider all the items in the Kerberos Conguration Checklist (page 152), in particular: examine the keytab le, with the following command:
klist -k <keytab>
Replace <keytab> with the path to your keytab le. check the congured hostname for your system, with the following command:
hostname -f
Ensure that this name matches the name in the keytab le, or use the saslHostName (page 1094) to pass MongoDB the correct hostname.
Enable the Traditional MongoDB Authentication Mechanism
For testing and development purposes you can enable both the Kerberos (i.e. GSSAPI) authentication mechanism in combination with the traditional MongoDB challenge/response authentication mechanism (i.e. MONGODB-CR), using the following setParameter (page 1087) run-time option:
mongod --setParameter authenticationMechanisms=GSSAPI,MONGODB-CR
Warning: All keyFile (page 1081) internal authentication between members of a replica set or sharded cluster still uses the MONGODB-CR authentication mechanism, even if MONGODB-CR is not enabled. All client authentication will still use Kerberos.
153
154
CHAPTER 20
Reference
20.1 User Privilege Roles in MongoDB

New in version 2.4. In version 2.4, MongoDB adds support for the following user roles:
20.1.1 Roles
Changed in version 2.4. Roles in MongoDB provide users with a set of specic privileges, on specic logical databases. Users may have multiple roles and may have different roles on different logical database. Roles only grant privileges and never limit access: if a user has read (page 155) and readWriteAnyDatabase permissions on the records database, that user will be able to write data to the records database. Note: By default, MongoDB 2.4 is backwards-compatible with the MongoDB 2.2 access control roles. You can explicitly disable this backwards-compatibility by setting the supportCompatibilityFormPrivilegeDocuments (page 1094) option to 0 during startup, as in the following command-line invocation of MongoDB:
mongod --setParameter supportCompatibilityFormPrivilegeDocuments=0
In general, you should set this option if your deployment does not need to support legacy user documents. Typically legacy user documents are only useful during the upgrade process and while you migrate applications to the updated privilege document form. See privilege documents (page 159) and Delegated Credentials for MongoDB Authentication (page 162) for more information about permissions and authentication in MongoDB. Database User Roles read Provides users with the ability to read data from any collection within a specic logical database. This includes find() (page 924) and the following database commands: aggregate (page 808) checkShardingIndex (page 848) cloneCollectionAsCapped (page 856) 155
collStats (page 873) count (page 808) dataSize (page 877) dbHash (page 877) dbStats (page 877) distinct (page 809) filemd5 (page 864) geoNear (page 822) geoSearch (page 823) geoWalk (page 823) group (page 810) mapReduce (page 814) (inline output only.) text (page 832) (beta feature.) readWrite Provides users with the ability to read from or write to any collection within a specic logical database. Users with readWrite (page 156) have access to all of the operations available to read (page 155) users, as well as the following basic write operations: insert() (page 936), remove() (page 944), and update() (page 948). Additionally, users with the readWrite (page 156) have access to the following database commands: cloneCollection (page 856) (as the target database.) convertToCapped (page 861) create (page 863) (and to create collections implicitly.) drop() (page 921) dropIndexes (page 864) emptycapped (page 915) ensureIndex() (page 921) findAndModify (page 826) mapReduce (page 814) (output to a collection.) renameCollection (page 867) (within the same database.) Database Administration Roles dbAdmin Provides the ability to perform the following set of administrative operations within the scope of this logical database. clean (page 856) collMod (page 857) collStats (page 873) compact (page 859)
156
Chapter 20. Reference
convertToCapped (page 861) create (page 863) db.createCollection() (page 970) dbStats (page 877) drop() (page 921) dropIndexes (page 864) ensureIndex() (page 921) indexStats (page 883) profile (page 888) reIndex (page 867) renameCollection (page 867) (within a single database.) validate (page 907) Furthermore, only dbAdmin (page 156) has the ability to read the system.profile (page 1099) collection. userAdmin Allows users to read and write data to the system.users (page 160) collection of any database. Users with this role will be able to modify permissions for existing users and create new users. userAdmin (page 157) does not restrict the permissions that a user can grant, and a userAdmin (page 157) user can grant privileges to themselves or other users in excess of the userAdmin (page 157) users current privileges. Important: userAdmin (page 157) is effectively the superuser role for a specic database. Users with userAdmin (page 157) can grant themselves all privileges. However, userAdmin (page 157) does not explicitly authorize a user for any privileges beyond user administration.
20.1.2 Administrative Roles

clusterAdmin clusterAdmin (page 157) grants access to several administration operations that affect or present information about the whole system, rather than just a single database. These privileges include but are not limited to replica set and sharded cluster administrative functions. clusterAdmin (page 157) is only applicable on the admin database. Specically, users with the clusterAdmin (page 157) role have access to the following operations: addShard (page 846) closeAllDatabases (page 857) connPoolStats (page 875) connPoolSync (page 861) _cpuProfilerStart _cpuProfilerStop cursorInfo (page 877) diagLogging (page 879) dropDatabase (page 863)
20.1. User Privilege Roles in MongoDB
157
enableSharding (page 848) flushRouterConfig (page 848) fsync (page 865) db.fsyncUnlock() (page 977) getCmdLineOpts (page 879) getLog (page 880) getParameter (page 866) getShardMap (page 848) getShardVersion (page 848) hostInfo (page 880) db.currentOp() (page 971) db.killOp() (page 981) listDatabases (page 888) listShards (page 849) logRotate (page 866) moveChunk (page 849) movePrimary (page 850) netstat (page 888) removeShard (page 850) repairDatabase (page 868) replSetFreeze (page 840) replSetGetStatus (page 840) replSetInitiate (page 842) replSetMaintenance (page 843) replSetReconfig (page 844) replSetStepDown (page 844) replSetSyncFrom (page 845) resync (page 846) serverStatus (page 889) setParameter (page 869) setShardVersion (page 851) shardCollection (page 851) shardingState (page 852) shutdown (page 869) splitChunk (page 852) splitVector (page 853)
158
split (page 853) top (page 906) touch (page 870) unsetSharding (page 855)
20.1.3 Any Database Roles

Note: You must specify the following any database roles on the admin databases. These roles apply to all databases in a mongod (page 1021) instance and are roughly equivalent to their single-database equivalents. If you add any of these roles to a user privilege document (page 159) outside of the admin database, the privilege will have no effect. readAnyDatabase readAnyDatabase (page 159) provides users with the same read-only permissions as read (page 155), except it applies to all logical databases in the MongoDB environment. readWriteAnyDatabase readWriteAnyDatabase (page 159) provides users with the same read and write permissions as readWrite (page 156), except it applies to all logical databases in the MongoDB environment. userAdminAnyDatabase userAdminAnyDatabase (page 159) provides users with the same access to user administration operations as userAdmin (page 157), except it applies to all logical databases in the MongoDB environment. Important: Because users with userAdminAnyDatabase (page 159) and userAdmin (page 157) have the ability to create and modify permissions in addition to their own level of access, this role is effectively the MongoDB system superuser. However, userAdminAnyDatabase (page 159) and userAdmin (page 157) do not explicitly authorize a user for any privileges beyond user administration. dbAdminAnyDatabase dbAdminAnyDatabase (page 159) provides users with the same access to database administration operations as dbAdmin (page 156), except it applies to all logical databases in the MongoDB environment.
20.1.4 Combined Access

Some operations are only available to users that have multiple roles. Consider the following: sh.status() (page 1002) Requires clusterAdmin (page 157) and read (page 155) access to the config (page 545) database. applyOps (page 837), eval (page 823) 1 Requires readWriteAnyDatabase (page 159), userAdminAnyDatabase (page 159), dbAdminAnyDatabase (page 159) and clusterAdmin (page 157) (on the admin database.)
20.2 system.users Privilege Documents

Changed in version 2.4.
20.2. system.users Privilege Documents
159
20.2.1 Overview
The documents in the <database>.system.users (page 160) collection store credentials and user privilege information used by the authentication system to provision access to users in the MongoDB system. See User Privilege Roles in MongoDB (page 155) for more information about access roles, and Security (page 129) for an overview security in MongoDB.
20.2.2 Data Model

<database>.system.users Changed in version 2.4. Documents in the <database>.system.users (page 160) collection stores credentials and user roles (page 155) for users who have access to the database. Consider the following prototypes of user privilege documents:
{ user: "<username>", pwd: "<hash>", roles: [] } { user: "<username>", userSource: "<database>", roles: [] }
Note: The pwd (page 160) and userSource (page 161) elds are mutually exclusive. A single document cannot contain both. The following privilege document with the otherDBRoles (page 161) eld is only supported on the admin database:
{ user: "<username>", userSource: "<database>", otherDBRoles: { <database0> : [], <database1> : [] }, roles: [] }
Consider the content of the following elds in the system.users (page 160) documents: <database>.system.users.user user (page 160) is a string that identies each user. Users exist in the context of a single logical database; however, users from one database may obtain access in another database by way of the otherDBRoles (page 161) eld on the admin database, the userSource (page 161) eld, or the Any Database Roles (page 159). <database>.system.users.pwd pwd (page 160) holds a hashed shared secret used to authenticate the user (page 160). pwd (page 160) eld is mutually exclusive with the userSource (page 161) eld.
160
<database>.system.users.roles roles (page 160) holds an array of user roles. The available roles are: read (page 155) readWrite (page 156) dbAdmin (page 156) userAdmin (page 157) clusterAdmin (page 157) readAnyDatabase (page 159) readWriteAnyDatabase (page 159) userAdminAnyDatabase (page 159) dbAdminAnyDatabase (page 159) See Roles (page 155) for full documentation of all available user roles. <database>.system.users.userSource A string that holds the name of the database that contains the credentials for the user. If userSource (page 161) is $external, then MongoDB will use an external resource, such as Kerberos, for authentication credentials. Note: In the current release, the only external authentication source is Kerberos, which is only available in MongoDB Enterprise. Use userSource (page 161) to ensure that a single users authentication credentials are only stored in a single location in a mongod (page 1021) instances data. A userSource (page 161) and user (page 160) pair identies a unique user in a MongoDB system. admin.system.users.otherDBRoles A document that holds one or more elds with a name that is the name of a database in the MongoDB instance with a value that holds a list of roles this user has on other databases. Consider the following example:
{ user: "admin", userSource: "$external", roles: [ "clusterAdmin"], otherDBRoles: { config: [ "read" ], records: [ "dbadmin" ] } }
This user has the following privileges: clusterAdmin (page 157) on the admin database, read (page 155) on the config (page 545) database, and dbAdmin (page 156) on the records database.
20.2. system.users Privilege Documents
161
20.2.3 Delegated Credentials for MongoDB Authentication

New in version 2.4. With a new document format in the system.users (page 160) collection, MongoDB now supports the ability to delegate authentication credentials to other sources and databases. The userSource (page 161) eld in these documents forces MongoDB to use another source for credentials. Consider the following document in a system.users (page 160) collection in a database named accounts:
{ user: "application0", pwd: "YvuolxMtaycghk2GMrzmImkG4073jzAw2AliMRul", roles: [] }
Then for every database that the application0 user requires access, add documents to the system.users (page 160) collection that resemble the following:
{ user: "application0", roles: [readWrite], userSource: "accounts" }
To gain privileges to databases where the application0 has access, you must rst authenticate to the accounts database.
20.2.4 Disable Legacy Privilege Documents

By default MongoDB 2.4 includes support for both new, role-based privilege documents style as well 2.2 and earlier privilege documents. MongoDB assumes any privilege document without a roles (page 160) eld is a 2.2 or earlier document. To ensure that mongod (page 1021) instances will only provide access to users dened with the new role-based privilege documents, use the following setParameter (page 1087) run-time option:
mongod --setParameter supportCompatibilityFormPrivilegeDocuments=0
20.3 Password Hashing Insecurity

In version 2.2 and earlier: the read-write users of a database all have access to the system.users collection, which contains the user names and user password hashes. 2 Note: In 2.4, only users with the userAdmin role have access to the system.users collection. if a user has the same password for multiple databases, the hash will be the same. A malicious user could exploit this to gain access on a second database using a different users credentials. As a result, always use unique username and password combinations for each database. Thanks to Will Urbanski, from Dell SecureWorks, for identifying this issue.
2
Read-only users do not have access to the system.users collection.
162
Part IV
Core MongoDB Operations (CRUD)
163
CRUD stands for create, read, update, and delete, which are the four core database operations used in database driven application development. The CRUD Operations for MongoDB (page 201) section provides introduction to each class of operation along with complete examples of each operation. The documents in the Read and Write Operations in MongoDB (page 167) section provide a higher level overview of the behavior and available functionality of these operations.
165
166
CHAPTER 21
Read and Write Operations in MongoDB
The Read Operations (page 167) and Write Operations (page 179) documents provide higher level introductions and description of the behavior and operations of read and write operations for MongoDB deployments. The BSON Documents (page 187) provides an overview of documents and document-orientation in MongoDB.
21.1 Read Operations

Read operations include all operations that return a cursor in response to application request data (i.e. queries,) and also include a number of aggregation (page 253) operations that do not return a cursor but have similar properties as queries. These commands include aggregate (page 808), count (page 808), and distinct (page 809). This document describes the syntax and structure of the queries applications use to request data from MongoDB and how different factors affect the efciency of reads. Note: All of the examples in this document use the mongo (page 1036) shell interface. All of these operations are available in an idiomatic interface for each language by way of the MongoDB Driver (page 555). See your driver documentation for full API documentation.
21.1.1 Queries in MongoDB

In the mongo (page 1036) shell, the find() (page 924) and findOne() (page 929) methods perform read operations. The find() (page 924) method has the following syntax: 1
db.collection.find( <query>, <projection> )
The db.collection object species the database and collection to query. All queries in MongoDB address a single collection. You can enter db in the mongo (page 1036) shell to return the name of the current database. Use the show collections operation in the mongo (page 1036) shell to list the current collections in the database. Queries in MongoDB are BSON objects that use a set of query operators (page 763) to describe query parameters. The <query> argument of the find() (page 924) method holds this query document. A read operation without a query document will return all documents in the collection.
1
db.collection.find() (page 924) is a wrapper for the more formal query structure with the $query operator.
167
The <projection> argument describes the result set in the form of a document. Projections specify or limit the elds to return. Without a projection, the operation will return all elds of the documents. Specify a projection if your documents are larger, or when your application only needs a subset of available elds. The order of documents returned by a query is not dened and is not necessarily consistent unless you specify a sort (sort() (page 965)). For example, the following operation on the inventory collection selects all documents where the type eld equals food and the price eld has a value less than 9.95. The projection limits the response to the item and qty, and _id eld:
db.inventory.find( { type: food, price: { $lt: 9.95 } }, { item: 1, qty: 1 } )
The findOne() (page 929) method is similar to the find() (page 924) method except the findOne() (page 929) method returns a single document from a collection rather than a cursor. The method has the syntax:
db.collection.findOne( <query>, <projection> )
For additional documentation and examples of the main MongoDB read operators, refer to the Read (page 209) page of the Core MongoDB Operations (CRUD) (page 165) section. Query Document This section provides an overview of the query document for MongoDB queries. See the preceding section for more information on queries in MongoDB (page 167). The following examples demonstrate the key properties of the query document in MongoDB queries, using the find() (page 924) method from the mongo (page 1036) shell, and a collection of documents named inventory: An empty query document ({}) selects all documents in the collection:
db.inventory.find( {} )
Not specifying a query document to the find() (page 924) is equivalent to specifying an empty query document. Therefore the following operation is equivalent to the previous operation:
db.inventory.find()
A single-clause query selects all documents in a collection where a eld has a certain value. These are simple equality queries. In the following example, the query selects all documents in the collection where the type eld has the value snacks:
db.inventory.find( { type: "snacks" } )
A single-clause query document can also select all documents in a collection given a condition or set of conditions for one eld in the collections documents. Use the query operators (page 763) to specify conditions in a MongoDB query. In the following example, the query selects all documents in the collection where the value of the type eld is either food or snacks:
db.inventory.find( { type: { $in: [ food, snacks ] } } )
Note: Although you can express this query using the $or (page 770) operator, choose the $in (page 765) operator rather than the $or (page 770) operator when performing equality checks on the same eld.
168
Chapter 21. Read and Write Operations in MongoDB
A compound query can specify conditions for more than one eld in the collections documents. Implicitly, a logical AND conjunction connects the clauses of a compound query so that the query selects the documents in the collection that match all the conditions. In the following example, the query document species an equality match on a single eld, followed by a range of values for a second eld using a comparison operator (page 763):
db.inventory.find( { type: food, price: { $lt: 9.95 } } )
This query selects all documents where the type eld has the value food and the value of the price eld is less than ($lt (page 766)) 9.95. Using the $or (page 770) operator, you can specify a compound query that joins each clause with a logical OR conjunction so that the query selects the documents in the collection that match at least one condition. In the following example, the query document selects all documents in the collection where the eld qty has a value greater than ($gt (page 764)) 100 or the value of the price eld is less than ($lt (page 766)) 9.95:
db.inventory.find( { $or: [ { qty: { $gt: 100 } }, { price: { $lt: 9.95 } } ] } )
With additional clauses, you can specify precise conditions for matching documents. In the following example, the compound query document selects all documents in the collection where the value of the type eld is food and either the qty has a value greater than ($gt (page 764)) 100 or the value of the price eld is less than ($lt (page 766)) 9.95:
db.inventory.find( { type: food, $or: [ { qty: { $gt: 100 } }, { price: { $lt: 9.95 } } ] } )
Subdocuments
When the eld holds an embedded document (i.e. subdocument), you can either specify the entire subdocument as the value of a eld, or reach into the subdocument using dot notation, to specify values for individual elds in the subdocument: Equality matches within subdocuments select documents if the subdocument matches exactly the specied subdocument, including the eld order. In the following example, the query matches all documents where the value of the eld producer is a subdocument that contains only the eld company with the value ABC123 and the eld address with the value 123 Street, in the exact order:
db.inventory.find( { producer: { company: ABC123, address: 123 Street } } )
Equality matches for specic elds within subdocuments select documents when the eld in the subdocument contains a eld that matches the specied value. In the following example, the query uses the dot notation to match all documents where the value of the eld producer is a subdocument that contains a eld company with the value ABC123 and may contain other elds: 21.1. Read Operations 169
db.inventory.find( { producer.company: ABC123 } )
Arrays
When the eld holds an array, you can query for values in the array, and if the array holds sub-documents, you query for specic elds within the sub-documents using dot notation: Equality matches can specify an entire array, to select an array that matches exactly. In the following example, the query matches all documents where the value of the eld tags is an array and holds three elements, fruit, food, and citrus, in this order:
db.inventory.find( { tags: [ fruit, food, citrus ] } )
Equality matches can specify a single element in the array. If the array contains at least one element with the specied value, as in the following example: the query matches all documents where the value of the eld tags is an array that contains, as one of its elements, the element fruit:
db.inventory.find( { tags: fruit } )
Equality matches can also select documents by values in an array using the array index (i.e. position) of the element in the array, as in the following example: the query uses the dot notation to match all documents where the value of the tags eld is an array whose rst element equals fruit:
db.inventory.find( { tags.0 : fruit } )
In the following examples, consider an array that contains subdocuments: If you know the array index of the subdocument, you can specify the document using the subdocuments position. The following example selects all documents where the memos contains an array whose rst element (i.e. index is 0) is a subdocument with the eld by with the value shipping:
db.inventory.find( { memos.0.by: shipping } )
If you do not know the index position of the subdocument, concatenate the name of the eld that contains the array, with a dot (.) and the name of the eld in the subdocument. The following example selects all documents where the memos eld contains an array that contains at least one subdocument with the eld by with the value shipping:
db.inventory.find( { memos.by: shipping } )
To match by multiple elds in the subdocument, you can use either dot notation or the $elemMatch (page 783) operator: The following example uses dot notation to query for documents where the value of the memos eld is an array that has at least one subdocument that contains the eld memo equal to on time and the eld by equal to shipping:
db.inventory.find( { memos.memo: on time, memos.by: shipping } )
170
The following example uses $elemMatch (page 783) to query for documents where the value of the memos eld is an array that has at least one subdocument that contains the eld memo equal to on time and the eld by equal to shipping:
db.inventory.find( { memos: { $elemMatch: { memo : on time, by: shipping } } } )
Refer to the Query, Update and Projection Operators (page 763) document for the complete list of query operators. Result Projections The projection specication limits the elds to return for all matching documents. Restricting the elds to return can minimize network transit costs and the costs of deserializing documents in the application layer. The second argument to the find() (page 924) method is a projection, and it takes the form of a document with a list of elds for inclusion or exclusion from the result set. You can either specify the elds to include (e.g. { field: 1 }) or specify the elds to exclude (e.g. { field: 0 }). The _id eld is, by default, included in the result set. To exclude the _id eld from the result set, you need to specify in the projection document the exclusion of the _id eld (i.e. { _id: 0 }). Note: You cannot combine inclusion and exclusion semantics in a single projection with the exception of the _id eld. Consider the following projection specications in find() (page 924) operations: If you specify no projection, the find() (page 924) method returns all elds of all documents that match the query.
db.inventory.find( { type: food } )
This operation will return all documents in the inventory collection where the value of the type eld is food. A projection can explicitly include several elds. In the following operation, find() (page 924) method returns all documents that match the query as well as item and qty elds. The results also include the _id eld:
db.inventory.find( { type: food }, { item: 1, qty: 1 } )
You can remove the _id eld from the results by specifying its exclusion in the projection, as in the following example:
db.inventory.find( { type: food }, { item: 1, qty: 1, _id:0 } )
This operation returns all documents that match the query, and only includes the item and qty elds in the result set. To exclude a single eld or group of elds you can use a projection in the following form:
db.inventory.find( { type: food }, { type:0 } )
This operation returns all documents where the value of the type eld is food, but does not include the type eld in the output. 21.1. Read Operations 171
With the exception of the _id eld you cannot combine inclusion and exclusion statements in projection documents. The $elemMatch (page 798) and $slice (page 801) projection operators provide more control when projecting only a portion of an array.
21.1.2 Indexes
Indexes improve the efciency of read operations by reducing the amount of data that query operations need to process and thereby simplifying the work associated with fullling queries within MongoDB. The indexes themselves are a special data structure that MongoDB maintains when inserting or modifying documents, and any given index can: support and optimize specic queries, sort operations, and allow for more efcient storage utilization. For more information about indexes in MongoDB see: Indexes (page 327) and Indexing Overview (page 329). You can create indexes using the db.collection.ensureIndex() (page 921) method in the mongo (page 1036) shell, as in the following prototype operation:
db.collection.ensureIndex( { <field1>: <order>, <field2>: <order>, ... } )
The field species the eld to index. The eld may be a eld from a subdocument, using dot notation to specify subdocument elds. You can create an index on a single eld or a compound index (page 331) that includes multiple elds in the index. The order option is species either ascending ( 1 ) or descending ( -1 ). MongoDB can read the index in either direction. In most cases, you only need to specify indexing order (page 332) to support sort operations in compound queries. Covering a Query An index covers (page 342) a query, a covered query, when: all the elds in the query (page 168) are part of that index, and all the elds returned in the documents that match the query are in the same index. For these queries, MongoDB does not need to inspect at documents outside of the index, which is often more efcient than inspecting entire documents. Example Given a collection inventory with the following index on the type and item elds:
{ type: 1, item: 1 }
This index will cover the following query on the type and item elds, which returns only the item eld:
db.inventory.find( { type: "food", item:/^c/ }, { item: 1, _id: 0 } )
However, this index will not cover the following query, which returns the item eld and the _id eld:
db.inventory.find( { type: "food", item:/^c/ }, { item: 1 } )
See Create Indexes that Support Covered Queries (page 342) for more information on the behavior and use of covered queries. 172 Chapter 21. Read and Write Operations in MongoDB
Measuring Index Use The explain() (page 953) cursor method allows you to inspect the operation of the query system, and is useful for analyzing the efciency of queries, and for determining how the query uses the index. Call the explain() (page 953) method on a cursor returned by find() (page 924), as in the following example:
db.inventory.find( { type: food } ).explain()
Note: Only use explain() (page 953) to test the query operation, and not the timing of query performance. Because explain() (page 953) attempts multiple query plans, it does not reect accurate query performance. If the above operation could not use an index, the output of explain() (page 953) would resemble the following:
{ "cursor" : "BasicCursor", "isMultiKey" : false, "n" : 5, "nscannedObjects" : 4000006, "nscanned" : 4000006, "nscannedObjectsAllPlans" : 4000006, "nscannedAllPlans" : 4000006, "scanAndOrder" : false, "indexOnly" : false, "nYields" : 2, "nChunkSkips" : 0, "millis" : 1591, "indexBounds" : { }, "server" : "mongodb0.example.net:27017" }
The BasicCursor value in the cursor (page 956) eld conrms that this query does not use an index. The explain.nscannedObjects (page 956) value shows that MongoDB must scan 4,000,006 documents to return only 5 documents. To increase the efciency of the query, create an index on the type eld, as in the following example:
db.inventory.ensureIndex( { type: 1 } )
Run the explain() (page 953) operation, as follows, to test the use of the index:
db.inventory.find( { type: food } ).explain()
Consider the results:

{ "cursor" : "BtreeCursor type_1", "isMultiKey" : false, "n" : 5, "nscannedObjects" : 5, "nscanned" : 5, "nscannedObjectsAllPlans" : 5, "nscannedAllPlans" : 5, "scanAndOrder" : false, "indexOnly" : false, "nYields" : 0, "nChunkSkips" : 0, "millis" : 0, "indexBounds" : { "type" : [ [ "food",
21.1. Read Operations
173
"food" ] ] }, "server" : "mongodbo0.example.net:27017" }
The BtreeCursor value of the cursor (page 956) eld indicates that the query used an index. This query: returned 5 documents, as indicated by the n (page 956) eld; scanned 5 documents from the index, as indicated by the nscanned (page 956) eld; then read 5 full documents from the collection, as indicated by the nscannedObjects (page 956) eld. Although the query uses an index to nd the matching documents, if indexOnly (page 956) is false then an index could not cover (page 172) the query: MongoDB could not both match the query conditions (page 168) and return the results using only this index. See Create Indexes that Support Covered Queries (page 342) for more information. Query Optimization The MongoDB query optimizer processes queries and chooses the most efcient query plan for a query given the available indexes. The query system then uses this query plan each time the query runs. The query optimizer occasionally reevaluates query plans as the content of the collection changes to ensure optimal query plans. To create a new query plan, the query optimizer: 1. runs the query against several candidate indexes in parallel. 2. records the matches in a common results buffer or buffers. If the candidate plans include only ordered query plans, there is a single common results buffer. If the candidate plans include only unordered query plans, there is a single common results buffer. If the candidate plans include both ordered query plans and unordered query plans, there are two common results buffers, one for the ordered plans and the other for the unordered plans. If an index returns a result already returned by another index, the optimizer skips the duplicate match. In the case of the two buffers, both buffers are de-duped. 3. stops the testing of candidate plans and selects an index when one of the following events occur: An unordered query plan has returned all the matching results; or An ordered query plan has returned all the matching results; or An ordered query plan has returned a threshold number of matching results: Version 2.0: Threshold is the query batch size. The default batch size is 101. Version 2.2: Threshold is 101. The selected index becomes the index specied in the query plan; future iterations of this query or queries with the same query pattern will use this index. Query pattern refers to query select conditions that differ only in the values, as in the following two queries with the same query pattern:
db.inventory.find( { type: food } ) db.inventory.find( { type: utensil } )
To manually compare the performance of a query using more than one index, you can use the hint() (page 958) and explain() (page 953) methods in conjunction, as in the following prototype:
db.collection.find().hint().explain()
174
The following operations each run the same query but will reect the use of the different indexes:
db.inventory.find( { type: food } ).hint( { type: 1 } ).explain() db.inventory.find( { type: food } ).hint( { type: 1, name: 1 }).explain()
This returns the statistics regarding the execution of the query. For more information on the output of explain() (page 953), see cursor.explain() (page 953). Note: If you run explain() (page 953) without including hint() (page 958), the query optimizer reevaluates the query and runs against multiple indexes before returning the query statistics. As collections change over time, the query optimizer deletes a query plan and reevaluates the after any of the following events: the collection receives 1,000 write operations. the reIndex (page 867) rebuilds the index. you add or drop an index. the mongod (page 1021) process restarts. For more information, see Indexing Strategies (page 341). Query Operations that Cannot Use Indexes Effectively Some query operations cannot use indexes effectively or cannot use indexes at all. Consider the following situations: The inequality operators $nin (page 767) and $ne (page 767) are not very selective, as they often match a large portion of the index. As a result, in most cases, a $nin (page 767) or $ne (page 767) query with an index may perform no better than a $nin (page 767) or $ne (page 767) query that must scan all documents in a collection. Queries that specify regular expressions, with inline JavaScript regular expressions or $regex (page 774) operator expressions, cannot use an index. However, the regular expression with anchors to the beginning of a string can use an index.
21.1.3 Cursors
The find() (page 924) method returns a cursor to the results; however, in the mongo (page 1036) shell, if the returned cursor is not assigned to a variable, then the cursor is automatically iterated up to 20 times 2 to print up to the rst 20 documents that match the query, as in the following example:
db.inventory.find( { type: food } );
When you assign the find() (page 924) to a variable: you can call the cursor variable in the shell to iterate up to 20 times 2 and print the matching documents, as in the following example:
var myCursor = db.inventory.find( { type: food } ); myCursor
you can use the cursor method next() (page 962) to access the documents, as in the following example:
2 You can use the DBQuery.shellBatchSize to change the number of iteration from the default value 20. See Executing Queries (page 584) for more information.
175
var myCursor = db.inventory.find( { type: food } ); var myDocument = myCursor.hasNext() ? myCursor.next() : null; if (myDocument) { var myItem = myDocument.item; print(tojson(myItem)); }
As an alternative print operation, consider the printjson() helper method to replace print(tojson()):
if (myDocument) { var myItem = myDocument.item; printjson(myItem); }
you can use the cursor method forEach() (page 958) to iterate the cursor and access the documents, as in the following example:
var myCursor = db.inventory.find( { type: food } );
myCursor.forEach(printjson);
See JavaScript cursor methods (page 951) and your driver (page 555) documentation for more information on cursor methods. Iterator Index In the mongo (page 1036) shell, you can use the toArray() (page 966) method to iterate the cursor and return the documents in an array, as in the following:
var myCursor = db.inventory.find( { type: food } ); var documentArray = myCursor.toArray(); var myDocument = documentArray[3];
The toArray() (page 966) method loads into RAM all documents returned by the cursor; the toArray() (page 966) method exhausts the cursor. Additionally, some drivers (page 555) provide access to the documents by using an index on the cursor (i.e. cursor[index]). This is a shortcut for rst calling the toArray() (page 966) method and then using an index on the resulting array. Consider the following example:
var myCursor = db.inventory.find( { type: food } ); var myDocument = myCursor[3];
The myCursor[3] is equivalent to the following example:

myCursor.toArray() [3];
Cursor Behaviors Consider the following behaviors related to cursors: By default, the server will automatically close the cursor after 10 minutes of inactivity or if client has exhausted the cursor. To override this behavior, you can specify the noTimeout wire protocol ag in your query; however, you should either close the cursor manually or exhaust the cursor. In the mongo (page 1036) shell, you can set the noTimeout ag:
176
var myCursor = db.inventory.find().addOption(DBQuery.Option.noTimeout);
See your driver (page 555) documentation for information on setting the noTimeout ag. See Cursor Flags (page 177) for a complete list of available cursor ags. Because the cursor is not isolated during its lifetime, intervening write operations may result in a cursor that returns a single document 3 more than once. To handle this situation, see the information on snapshot mode (page 721). The MongoDB server returns the query results in batches: For most queries, the rst batch returns 101 documents or just enough documents to exceed 1 megabyte. Subsequent batch size is 4 megabytes. To override the default size of the batch, see batchSize() (page 952) and limit() (page 959). For queries that include a sort operation without an index, the server must load all the documents in memory to perform the sort and will return all documents in the rst batch. Batch size will not exceed the maximum BSON document size (page 1105). As you iterate through the cursor and reach the end of the returned batch, if there are more results, cursor.next() (page 962) will perform a getmore operation (page 973) to retrieve the next batch. To see how many documents remain in the batch as you iterate the cursor, you can use the objsLeftInBatch() (page 963) method, as in the following example:
var myCursor = db.inventory.find(); var myFirstDocument = myCursor.hasNext() ? myCursor.next() : null; myCursor.objsLeftInBatch();
You can use the command cursorInfo (page 877) to retrieve the following information on cursors: total number of open cursors size of the client cursors in current use number of timed out cursors since the last server restart Consider the following example:
db.runCommand( { cursorInfo: 1 } )
The result from the command returns the following document:

{ "totalOpen" : <number>, "clientCursors_size" : <number>, "timedOut" : <number>, "ok" : 1 }
Cursor Flags The mongo (page 1036) shell provides the following cursor ags: DBQuery.Option.tailable
A single document relative to value of the _id eld. A cursor cannot return the same document more than once if the document has not changed.
3
177
DBQuery.Option.slaveOk DBQuery.Option.oplogReplay DBQuery.Option.noTimeout DBQuery.Option.awaitData DBQuery.Option.exhaust DBQuery.Option.partial Aggregation Changed in version 2.2. MongoDB can perform some basic data aggregation operations on results before returning data to the application. These operations are not queries; they use database commands rather than queries, and they do not return a cursor. However, they still require MongoDB to read data. Running aggregation operations on the database side can be more efcient than running them in the application layer and can reduce the amount of data MongoDB needs to send to the application. These aggregation operations include basic grouping, counting, and even processing data using a map reduce framework. Additionally, in 2.2 MongoDB provides a complete aggregation framework for more rich aggregation operations. The aggregation framework provides users with a pipeline like framework: documents enter from a collection and then pass through a series of steps by a sequence of pipeline operators (page 291) that manipulate and transform the documents until theyre output at the end. The aggregation framework is accessible via the aggregate (page 808) command or the db.collection.aggregate() (page 918) helper in the mongo (page 1036) shell. For more information on the aggregation framework see Aggregation (page 253). Additionally, MongoDB provides a number of simple data aggregation operations for more basic data aggregation operations: count (page 808) (count() (page 952)) distinct (page 809) (db.collection.distinct() (page 920)) group (page 810) (db.collection.group() (page 932)) mapReduce (page 814). (Also consider mapReduce() (page 937) and Map-Reduce (page 311).)
21.1.4 Architecture
Read Operations from Sharded Clusters Sharded clusters allow you to partition a data set among a cluster of mongod (page 1021) in a way that is nearly transparent to the application. See the Sharding (page 483) section of this manual for additional information about these deployments. For a sharded cluster, you issue all operations to one of the mongos (page 1032) instances associated with the cluster. mongos (page 1032) instances route operations to the mongod (page 1021) in the cluster and behave like mongod (page 1021) instances to the application. Read operations to a sharded collection in a sharded cluster are largely the same as operations to a replica set or standalone instances. See the section on Read Operations in Sharded Clusters (page 490) for more information. In sharded deployments, the mongos (page 1032) instance routes the queries from the clients to the mongod (page 1021) instances that hold the data, using the cluster metadata stored in the cong database (page 500).
178
For sharded collections, if queries do not include the shard key (page 485), the mongos (page 1032) must direct the query to all shards in a collection. These scatter gather queries can be inefcient, particularly on larger clusters, and are unfeasible for routine operations. For more information on read operations in sharded clusters, consider the following resources: An Introduction to Shard Keys (page 485) Shard Key Internals and Operations (page 493) Querying Sharded Clusters (page 494) mongos Operational Overview (page 490) Read Operations from Replica Sets Replica sets use read preferences to determine where and how to route read operations to members of the replica set. By default, MongoDB always reads data from a replica sets primary. You can modify that behavior by changing the read preference mode (page 402). You can congure the read preference mode (page 402) on a per-connection or per-operation basis to allow reads from secondaries to: reduce latency in multi-data-center deployments, improve read throughput by distributing high read-volumes (relative to write volume), for backup operations, and/or to allow reads during failover (page 389) situations. Read operations from secondary members of replica sets are not guaranteed to reect the current state of the primary, and the state of secondaries will trail the primary by some amount of time. Often, applications dont rely on this kind of strict consistency, but application developers should always consider the needs of their application before setting read preference. For more information on read preference or on the read preference modes, see Read Preference (page 402) and Read Preference Modes (page 402).
21.2 Write Operations

All operations that create or modify data in the MongoDB instance are write operations. MongoDB represents data as BSON documents stored in collections. Write operations target one collection and are atomic on the level of a single document: no single write operation can atomically affect more than one document or more than one collection. This document introduces the write operators available in MongoDB as well as presents strategies to increase the efciency of writes in applications.
21.2.1 Write Operators

For information on write operators and how to write data to a MongoDB database, see the following pages: Create (page 201) Update (page 219) Delete (page 227)
21.2. Write Operations
179
For information on specic methods used to perform write operations in the mongo (page 1036) shell, see the following: db.collection.insert() (page 936) db.collection.update() (page 948) db.collection.save() (page 946) db.collection.findAndModify() (page 926) db.collection.remove() (page 944) For information on how to perform write operations from within an application, see the MongoDB Drivers and Client Libraries (page 555) documentation or the documentation for your client library.
21.2.2 Write Concern

Write concern is a quality of every write operation issued to a MongoDB deployment, and describes the amount of concern the application has for the outcome of the write operation. With weak or disabled write concern, the application can send an write operation to MongoDB and then continue without waiting for a response from the database. With stronger write concerns, write operations wait until MongoDB acknowledges or conrms a successful write operation. MongoDB provides different levels of write concern to better address the specic needs of applications. Note: The driver write concern (page 1185) change created a new connection class in all of the MongoDB drivers, called MongoClient with a different default write concern. See the release notes (page 1185) for this change, and the release notes for the driver youre using for more information about your drivers release. See also: Write Concern (page 398) and Write Concern Reference (page 183).
21.2.3 Bulk Inserts

In some situations you may need to insert or ingest a large amount of data into a MongoDB database. These bulk inserts have some special considerations that are different from other write operations. The insert() (page 936) method, when passed an array of documents, will perform a bulk insert, and inserts each document atomically. Drivers (page 555) provide their own interface for this kind of operation. New in version 2.2: insert() (page 936) in the mongo (page 1036) shell gained support for bulk inserts in version 2.2. Bulk insert can signicantly increase performance by amortizing write concern (page 180) costs. In the drivers, you can congure write concern for batches rather than on a per-document level. Drivers also have a ContinueOnError option in their insert operation, so that the bulk operation will continue to insert remaining documents in a batch even if an insert fails. Note: New in version 2.0: Support for ContinueOnError depends on version 2.0 of the core mongod (page 1021) and mongos (page 1032) components. If the bulk insert process generates more than one error in a batch job, the client will only receive the most recent error. All bulk operations to a sharded collection run with ContinueOnError, which applications cannot disable. See Strategies for Bulk Inserts in Sharded Clusters (page 525) section for more information on consideration for bulk inserts in sharded clusters.
180
For more information see your driver documentation (page 555) for details on performing bulk inserts in your application. Also consider the following resources: Sharded Clusters (page 183), Strategies for Bulk Inserts in Sharded Clusters (page 525), and Import and Export MongoDB Data (page 105).
21.2.4 Indexing
After every insert, update, or delete operation, MongoDB must update every index associated with the collection in addition to the data itself. Therefore, every index on a collection adds some amount of overhead for the performance of write operations. 4 In general, the performance gains that indexes provide for read operations are worth the insertion penalty; however, when optimizing write performance, be careful when creating new indexes and always evaluate the indexes on the collection and ensure that your queries are actually using these indexes. For more information on indexes in MongoDB consider Indexes (page 327) and Indexing Strategies (page 341).
21.2.5 Isolation
When a single write operation modies multiple documents, the operation as a whole is not atomic, and other operations may interleave. The modication of a single document, or record, is always atomic, even if the write operation modies multiple sub-document within the single record. No other operations are atomic; however, you can attempt to isolate a write operation that affects multiple documents using the isolation operator (page 796). To isolate a sequence of write operations from other read and write operations, see Perform Two Phase Commits (page 563).
21.2.6 Updates
Each document in a MongoDB collection has allocated record space which includes the entire document and a small amount of padding. This padding makes it possible for update operations to increase the size of a document slightly without causing the document to outgrow the allocated record size. Documents in MongoDB can grow up to the full maximum BSON document size (page 1105). However, when documents outgrow their allocated record size MongoDB must allocate a new record and move the document to the new record. Update operations that do not cause a document to grow, (i.e. in-place updates,) are signicantly more efcient than those updates that cause document growth. Use data models (page 233) that minimize the need for document growth when possible. For complete examples of update operations, see Update (page 219).
21.2.7 Padding Factor

If an update operation does not cause the document to increase in size, MongoDB can apply the update in-place. Some updates change the size of the document, for example using the $push (page 792) operator to append a sub-document to an array can cause the top level document to grow beyond its allocated space. When documents grow, MongoDB relocates the document on disk with enough contiguous space to hold the document. These relocations take longer than in-place updates, particularly if the collection has indexes that MongoDB must update all index entries. If collection has many indexes, the move will impact write throughput.
4 The overhead for sparse indexes (page 334) inserts and updates to un-indexed elds is less than for non-sparse indexes. Also for non-sparse indexes, updates that dont change the record size have less indexing overhead.
21.2. Write Operations
181
To minimize document movements, MongoDB employs padding. MongoDB adaptively learns if documents in a collection tend to grow, and if they do, adds a paddingFactor (page 874) so that the documents have room to grow on subsequent writes. The paddingFactor (page 874) indicates the padding for new inserts and moves. New in version 2.2: You can use the collMod (page 857) command with the usePowerOf2Sizes (page 857) ag so that MongoDB allocates document space in sizes that are powers of 2. This helps ensure that MongoDB can efciently reuse the space freed as a result of deletions or document relocations. As with all padding, using document space allocations with power of 2 sizes minimizes, but does not eliminate, document movements. To check the current paddingFactor (page 874) on a collection, you can run the db.collection.stats() (page 947) operation in the mongo (page 1036) shell, as in the following example:
db.myCollection.stats()
Since MongoDB writes each document at a different point in time, the padding for each document will not be the same. You can calculate the padding size by subtracting 1 from the paddingFactor (page 874), for example:
padding size = (paddingFactor - 1) * <document size>.
For example, a paddingFactor (page 874) of 1.0 species no padding whereas a paddingFactor of 1.5 species a padding size of 0.5 or 50 percent (50%) of the document size. Because the paddingFactor (page 874) is relative to the size of each document, you cannot calculate the exact amount of padding for a collection based on the average document size and padding factor. If an update operation causes the document to decrease in size, for instance if you perform an $unset (page 788) or a $pop (page 790) update, the document remains in place and effectively has more padding. If the document remains this size, the space is not reclaimed until you perform a compact (page 859) or a repairDatabase (page 868) operation. Note: The following operations remove padding: compact (page 859), repairDatabase (page 868), and initial replica sync operations. However, with the compact (page 859) command, you can run the command with a paddingFactor or a paddingBytes parameter. Padding is also removed if you use mongoexport (page 1059) from a collection. If you use mongoimport (page 1056) into a new collection, mongoimport (page 1056) will not add padding. If you use mongoimport (page 1056) with an existing collection with padding, mongoimport (page 1056) will not affect the existing padding. When a database operation removes padding, subsequent update that require changes in record sizes will have reduced throughput until the collections padding factor grows. Padding does not affect in-place, and after compact (page 859), repairDatabase (page 868), and replica set initial sync the collection will require less storage. See also: Can I manually pad documents to prevent moves during updates? (page 722) The $inc (page 784) for in-place updates.
182
21.2.8 Architecture
Replica Sets In replica sets, all write operations go to the sets primary, which applies the write operation then records the operations on the primarys operation log or oplog. The oplog is a reproducible sequence of operations to the data set. Secondary members of the set are continuously replicating the oplog and applying the operations to themselves in an asynchronous process. Large volumes of write operations, particularly bulk operations, may create situations where the secondary members have difculty applying the replicating operations from the primary at a sufcient rate: this can cause the secondarys state to fall behind that of the primary. Secondaries that are signicantly behind the primary present problems for normal operation of the replica set, particularly failover (page 389) in the form of rollbacks (page 390) as well as general read consistency (page 390). To help avoid this issue, you can customize the write concern (page 180) to return conrmation of the write operation to another member 5 of the replica set every 100 or 1,000 operations. This provides an opportunity for secondaries to catch up with the primary. Write concern can slow the overall progress of write operations but ensure that the secondaries can maintain a largely current state with respect to the primary. For more information on replica sets and write operations, see Replica Acknowledged (page 399), Oplog (page 392), Oplog Internals (page 408), and Change the Size of the Oplog (page 435). Sharded Clusters In a sharded cluster, MongoDB directs a given write operation to a shard and then performs the write on a particular chunk on that shard. Shards and chunks are range-based. Shard keys affect how MongoDB distributes documents among shards. Choosing the correct shard key can have a great impact on the performance, capability, and functioning of your database and cluster. For more information, see Sharded Cluster Administration (page 503) and Bulk Inserts (page 180).
21.3 Write Concern Reference

21.3.1 Overview
Write concern is a quality of every write operation issued to a MongoDB deployment, and describes the amount of concern the application has for the outcome of the write operation. With weak or disabled write concern, the application can send an write operation to MongoDB and then continue without waiting for a response from the database. With stronger write concerns, write operations wait until MongoDB acknowledges or conrms a successful write operation. MongoDB provides different levels of write concern to better address the specic needs of applications. See also: Write Concern (page 398) for an introduction to write concern in MongoDB.
21.3.2 Available Write Concern

To provide write concern, drivers (page 555) issue the getLastError (page 831) command after a write operation and receive a document with information about the last operation. This documents err eld contains either:
Calling getLastError (page 831) intermittently with a w value of 2 or majority will slow the throughput of write trafc; however, this practice will allow the secondaries to remain current with the state of the primary.
5
21.3. Write Concern Reference
183
null, which indicates the write operations have completed successfully, or a description of the last error encountered. The denition of a successful write depends on the arguments specied to getLastError (page 831), or in replica sets, the conguration of getLastErrorDefaults (page 466). When deciding the level of write concern for your application, see the introduction to Write Concern (page 398). The getLastError (page 831) command has the following options to congure write concern requirements: j or journal option This option conrms that the mongod (page 1021) instance has written the data to the on-disk journal and ensures data is not lost if the mongod (page 1021) instance shuts down unexpectedly. Set to true to enable, as shown in the following example:
db.runCommand( { getLastError: 1, j: "true" } )
If you set journal (page 1083) to true, and the mongod (page 1021) does not have journaling enabled, as with nojournal (page 1084), then getLastError (page 831) will provide basic receipt acknowledgment, and will include a jnote eld in its return document. w option This option provides the ability to disable write concern entirely as well as species the write concern operations for replica sets. See Write Concern Considerations (page 398) for an introduction to the fundamental concepts of write concern. By default, the w option is set to 1, which provides basic receipt acknowledgment on a single mongod (page 1021) instance or on the primary in a replica set. The w option takes the following values: -1: Disables all acknowledgment of write operations, and suppresses all errors, including network and socket errors. 0: Disables basic acknowledgment of write operations, but returns information about socket exceptions and networking errors to the application. Note: If you disable basic write operation acknowledgment but require journal commit acknowledgment, the journal commit prevails, and the driver will require that mongod (page 1021) will acknowledge the write operation. 1: Provides acknowledgment of write operations on a standalone mongod (page 1021) or the primary in a replica set. A number greater than 1: Guarantees that write operations have propagated successfully to the specied number of replica set members including the primary. If you set w to a number that is greater than the number of set members that hold data, MongoDB waits for the non-existent members to become available, which means MongoDB blocks indenitely. majority: Conrms that write operations have propagated to the majority of congured replica set: a majority of the sets congured members must acknowledge the write operation before it succeeds. This ensures that write operation will never be subject to a rollback in the course of normal operation, and furthermore allows you to avoid hard coding assumptions about the size of your replica set into your application. 184 Chapter 21. Read and Write Operations in MongoDB
A tag set: By specifying a tag set (page 455) you can have ne-grained control over which replica set members must acknowledge a write operation to satisfy the required level of write concern. getLastError (page 831) also supports a wtimeout setting which allows clients to specify a timeout for the write concern: if you dont specify wtimeout and the mongod (page 1021) cannot fulll the write concern the getLastError (page 831) will block, potentially forever. For more information on write concern and replica sets, see Write Concern for Replica Sets (page 399) for more information. In sharded clusters, mongos (page 1032) instances will pass write concern on to the shard mongod (page 1021) instances.
21.3. Write Concern Reference
185
186
CHAPTER 22
Fundamental Concepts for Document Databases
22.1 BSON Documents

MongoDB is a document-based database system, and as a result, all records, or data, in MongoDB are documents. Documents are the default representation of most user accessible data structures in the database. Documents provide structure for data in the following MongoDB contexts: the records (page 189) stored in collections the query selectors (page 190) that determine which records to select for read, update, and delete operations the update actions (page 191) that specify the particular eld updates to perform during an update operation the specication of indexes (page 192) for collection. arguments to several MongoDB methods and operators, including: sort order (page 192) for the sort() (page 965) method. index specication (page 192) for the hint() (page 958) method. the output of a number of MongoDB commands and operations, including: the output of collStats (page 873) command, and the output (page 889) of the serverStatus (page 889) command.
22.1.1 Structure
The document structure in MongoDB are BSON objects with support for the full range of BSON types; however, BSON documents are conceptually, similar to JSON objects, and have the following structure:
{ field1: field2: field3: ... fieldN: } value1, value2, value3, valueN
Having support for the full range of BSON types, MongoDB documents may contain eld and value pairs where the value can be another document, an array, an array of documents as well as the basic types such as Double, String, and Date. See also BSON Type Considerations (page 192). Consider the following document that contains values of varying types:
187
var mydoc = { _id: ObjectId("5099803df3f4948bd2f98391"), name: { first: "Alan", last: "Turing" }, birth: new Date(Jun 23, 1912), death: new Date(Jun 07, 1954), contribs: [ "Turing machine", "Turing test", "Turingery" ], views : NumberLong(1250000) }
The document contains the following elds: _id that holds an ObjectId. name that holds a subdocument that contains the elds first and last. birth and death, which both have Date types. contribs that holds an array of strings. views that holds a value of NumberLong type. All eld names are strings in BSON documents. Be aware that there are some restrictions on field names (page 1108) for BSON documents: eld names cannot contain null characters, dots (.), or dollar signs ($). Note: BSON documents may have more than one eld with the same name; however, most MongoDB Interfaces (page 555) represent MongoDB with a structure (e.g. a hash table) that does not support duplicate eld names. If you need to manipulate documents that have more than one eld with the same name, see your drivers documentation for more information. Some documents created by internal MongoDB processes may have duplicate elds, but no MongoDB process will ever add duplicate keys to an existing user document.
Type Operators To determine the type of elds, the mongo (page 1036) shell provides the following operators: instanceof returns a boolean to test if a value has a specic type. typeof returns the type of a eld. Example Consider the following operations using instanceof and typeof: The following operation tests whether the _id eld is of type ObjectId:
mydoc._id instanceof ObjectId
The operation returns true. The following operation returns the type of the _id eld:
typeof mydoc._id
In this case typeof will return the more generic object type rather than ObjectId type.
Dot Notation MongoDB uses the dot notation to access the elements of an array and to access the elds of a subdocument. 188 Chapter 22. Fundamental Concepts for Document Databases
To access an element of an array by the zero-based index position, you concatenate the array name with the dot (.) and zero-based index position:
<array>.<index>
To access a eld of a subdocument with dot-notation, you concatenate the subdocument name with the dot (.) and the eld name:
<subdocument>.<field>
See also: Subdocuments (page 169) for dot notation examples with subdocuments. Arrays (page 170) for dot notation examples with arrays.
22.1.2 Document Types in MongoDB

Record Documents Most documents in MongoDB in collections store data from users applications. These documents have the following attributes: The maximum BSON document size is 16 megabytes. The maximum document size helps ensure that a single document cannot use excessive amount of RAM or, during transmission, excessive amount of bandwidth. To store documents larger than the maximum size, MongoDB provides the GridFS API. See mongofiles (page 1075) and the documentation for your driver (page 555) for more information about GridFS. Documents (page 187) have the following restrictions on eld names: The eld name _id is reserved for use as a primary key; its value must be unique in the collection, is immutable, and may be of any type other than an array. The eld names cannot start with the $ character. The eld names cannot contain the . character. Note: Most MongoDB driver clients will include the _id eld and generate an ObjectId before sending the insert operation to MongoDB; however, if the client sends a document without an _id eld, the mongod (page 1021) will add the _id eld and generate the ObjectId. The following document species a record in a collection:
{ _id: 1, name: { first: John, last: Backus }, birth: new Date(Dec 03, 1924), death: new Date(Mar 17, 2007), contribs: [ Fortran, ALGOL, Backus-Naur Form, FP ], awards: [ { award: National Medal of Science, year: 1975, by: National Science Foundation }, { award: Turing Award, year: 1977, by: ACM }
22.1. BSON Documents
189
] }
The document contains the following elds: _id, which must hold a unique value and is immutable. name that holds another document. This sub-document contains the elds first and last, which both hold strings. birth and death that both have date types. contribs that holds an array of strings. awards that holds an array of documents. Consider the following behavior and constraints of the _id eld in MongoDB documents: In documents, the _id eld is always indexed for regular collections. The _id eld may contain values of any BSON data type other than an array. Warning: To ensure functioning replication, do not store values that are of the BSON regular expression type in the _id eld. Consider the following options for the value of an _id eld: Use an ObjectId. See the ObjectId (page 194) documentation. Although it is common to assign ObjectId values to _id elds, if your objects have a natural unique identier, consider using that for the value of _id to save space and to avoid an additional index. Generate a sequence number for the documents in your collection in your application and use this value for the _id value. See the Create an Auto-Incrementing Sequence Field (page 572) tutorial for an implementation pattern. Generate a UUID in your application code. For a more efcient storage of the UUID values in the collection and in the _id index, store the UUID as a value of the BSON BinData type. Index keys that are of the BinData type are more efciently stored in the index if: the binary subtype value is in the range of 0-7 or 128-135, and the length of the byte array is: 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, or 32. Use your drivers BSON UUID facility to generate UUIDs. Be aware that driver implementations may implement UUID serialization and deserialization logic differently, which may not be fully compatible with other drivers. See your driver documentation for information concerning UUID interoperability. Query Specication Documents Query documents specify the conditions that determine which records to select for read, update, and delete operations. You can use <field>:<value> expressions to specify the equality condition and query operator (page 763) expressions to specifyo additional conditions. When passed as an argument to methods such as the find() (page 924) method, the remove() (page 944) method, or the update() (page 948) method, the query document selects documents for MongoDB to return, remove, or update, as in the following:
190
Chapter 22. Fundamental Concepts for Document Databases
db.bios.find( { _id: 1 } ) db.bios.remove( { _id: { $gt: 3 } } ) db.bios.update( { _id: 1, name: { first: John, last: Backus } }, <update>, <options> )
See also: Query Document (page 168) and Read (page 209) for more examples on selecting documents for reads. Update (page 219) for more examples on selecting documents for updates. Delete (page 227) for more examples on selecting documents for deletes. Update Specication Documents Update documents specify the data modications to perform during an update() (page 948) operation to modify existing records in a collection. You can use update operators (page 784) to specify the exact actions to perform on the document elds. Consider the update document example:
{ $set: { name.middle: Warner }, $push: { awards: { award: IBM Fellow, year: 1963, by: IBM } } }
When passed as an argument to the update() (page 948) method, the update actions document: Modies the eld name whose value is another document. Specically, the $set (page 788) operator updates the middle eld in the name subdocument. The document uses dot notation (page 188) to access a eld in a subdocument. Adds an element to the eld awards whose value is an array. Specically, the $push (page 792) operator adds another document as element to the eld awards.
db.bios.update( { _id: 1 }, { $set: { name.middle: Warner }, $push: { awards: { award: IBM Fellow, year: 1963, by: IBM } } } )
See also: update operators (page 784) page for the available update operators and syntax. update (page 219) for more examples on update documents. For additional examples of updates that involve array elements, including where the elements are documents, see the $ (page 788) positional operator.
191
Index Specication Documents Index specication documents describe the elds to index on during the index creation (page 921). See indexes (page 329) for an overview of indexes. 1 Index documents contain eld and value pairs, in the following form:
{ field: value }
field is the eld in the documents to index. value is either 1 for ascending or -1 for descending. The following document species the multi-key index (page 332) on the _id eld and the last eld contained in the subdocument name eld. The document uses dot notation (page 188) to access a eld in a subdocument:
{ _id: 1, name.last: 1 }
When passed as an argument to the ensureIndex() (page 921) method, the index documents species the index to create:
db.bios.ensureIndex( { _id: 1, name.last: 1 } )
Sort Order Specication Documents Sort order documents specify the order of documents that a query() (page 924) returns. Pass sort order specication documents as an argument to the sort() (page 965) method. See the sort() (page 965) page for more information on sorting. The sort order documents contain eld and value pairs, in the following form:
{ field: value }
field is the eld by which to sort documents. value is either 1 for ascending or -1 for descending. The following document species the sort order using the elds from a sub-document name rst sort by the last eld ascending, then by the first eld also ascending:
{ name.last: 1, name.first: 1 }
When passed as an argument to the sort() (page 965) method, the sort order document sorts the results of the find() (page 924) method:
db.bios.find().sort( { name.last: 1, name.first: 1 } )
22.1.3 BSON Type Considerations

The following BSON types require special consideration: ObjectId ObjectIds are: small, likely unique, fast to generate, and ordered. These values consists of 12-bytes, where the rst 4-bytes is a timestamp that reects the ObjectIds creation. Refer to the ObjectId (page 194) documentation for more information.
1
Indexes optimize a number of key read (page 167) and write (page 179) operations.
192
String BSON strings are UTF-8. In general, drivers for each programming language convert from the languages string format to UTF-8 when serializing and deserializing BSON. This makes it possible to store most international characters in BSON strings with ease. 2 In addition, MongoDB $regex (page 774) queries support UTF-8 in the regex string. Timestamps BSON has a special timestamp type for internal MongoDB use and is not associated with the regular Date (page 193) type. Timestamp values are a 64 bit value where: the rst 32 bits are a time_t value (seconds since the Unix epoch) the second 32 bits are an incrementing ordinal for operations within a given second. Within a single mongod (page 1021) instance, timestamp values are always unique. In replication, the oplog has a ts eld. The values in this eld reect the operation time, which uses a BSON timestamp value. Note: The BSON Timestamp type is for internal MongoDB use. For most cases, in application development, you will want to use the BSON date type. See Date (page 193) for more information. If you create a BSON Timestamp using the empty constructor (e.g. new Timestamp()), MongoDB will only generate a timestamp if you use the constructor in the rst eld of the document. 3 Otherwise, MongoDB will generate an empty timestamp value (i.e. Timestamp(0, 0).) Changed in version 2.1: mongo (page 1036) shell displays the Timestamp value with the wrapper:
Timestamp(<time_t>, <ordinal>)
Prior to version 2.1, the mongo (page 1036) shell display the Timestamp value as a document:
{ t : <time_t>, i : <ordinal> }
Date BSON Date is a 64-bit integer that represents the number of milliseconds since the Unix epoch (Jan 1, 1970). The ofcial BSON specication refers to the BSON Date type as the UTC datetime. Changed in version 2.0: BSON Date type is signed. Consider the following examples of BSON Date: Construct a Date using the new Date() constructor in the mongo (page 1036) shell:
var mydate1 = new Date()
4
Negative values represent dates before 1970.
Construct a Date using the ISODate() constructor in the mongo (page 1036) shell:
var mydate2 = ISODate()
2 Given strings using UTF-8 character sets, using sort() (page 965) on strings will be reasonably correct; however, because internally sort() (page 965) uses the C++ strcmp api, the sort order may handle some characters incorrectly. 3 If the rst eld in the document is _id, then you can generate a timestamp in the second eld of a document. 4 Prior to version 2.0, Date values were incorrectly interpreted as unsigned integers, which affected sorts, range queries, and indexes on Date elds. Because indexes are not recreated when upgrading, please re-index if you created an index on Date values with an earlier version, and dates before 1970 are relevant to your application.
193
Return the Date value as string:

mydate1.toString()
Return the month portion of the Date value; months are zero-indexed, so that January is month 0:
mydate1.getMonth()
22.2 ObjectId
22.2.1 Overview
ObjectId is a 12-byte BSON type, constructed using: a 4-byte value representing the seconds since the Unix epoch, a 3-byte machine identier, a 2-byte process id, and a 3-byte counter, starting with a random value. In MongoDB, documents stored in a collection require a unique _id eld that acts as a primary key. Because ObjectIds are small, most likely unique, and fast to generate, MongoDB uses ObjectIds as the default value for the _id eld if the _id eld is not specied; i.e., the mongod (page 1021) adds the _id eld and generates a unique ObjectId to assign as its value. Using ObjectIds for the _id eld, provides the following additional benets: in the mongo (page 1036) shell, you can access the creation time of the ObjectId, using the getTimestamp() method. sorting on an _id eld that stores ObjectId values is roughly equivalent to sorting by creation time. Important: The relationship between the order of ObjectId values and generation time is not strict within a single second. If multiple systems, or multiple processes or threads on a single system generate values, within a single second; ObjectId values do not represent a strict insertion order. Clock skew between clients can also result in non-strict ordering even for values, because client drivers generate ObjectId values, not the mongod (page 1021) process. Also consider the BSON Documents (page 187) section for related information on MongoDBs document orientation.
22.2.2 ObjectId()
The mongo (page 1036) shell provides the ObjectId() wrapper class to generate a new ObjectId, and to provide the following helper attribute and methods: str The hexadecimal string value of the ObjectId() object. getTimestamp() Returns the timestamp portion of the ObjectId() object as a Date. toString()
194
Returns the string representation of the ObjectId() object. The returned string literal has the format ObjectId(...). Changed in version 2.2: In previous versions ObjectId.toString() returns the value of the ObjectId as a hexadecimal string. valueOf() Returns the value of the ObjectId() object as a hexadecimal string. The returned string is the str attribute. Changed in version 2.2: In previous versions ObjectId.valueOf() returns the ObjectId() object.
22.2.3 Examples
Consider the following uses ObjectId() class in the mongo (page 1036) shell: To generate a new ObjectId, use the ObjectId() constructor with no argument:
x = ObjectId()
In this example, the value of x would be:

ObjectId("507f1f77bcf86cd799439011")
To generate a new ObjectId using the ObjectId() constructor with a unique hexadecimal string:
y = ObjectId("507f191e810c19729de860ea")
In this example, the value of y would be:

ObjectId("507f191e810c19729de860ea")
To return the timestamp of an ObjectId() object, use the getTimestamp() method as follows:
ObjectId("507f191e810c19729de860ea").getTimestamp()
This operation will return the following Date object:

ISODate("2012-10-17T20:46:22Z")
Access the str attribute of an ObjectId() object, as follows:

ObjectId("507f191e810c19729de860ea").str
This operation will return the following hexadecimal string:

507f191e810c19729de860ea
To return the string representation of an ObjectId() object, use the toString() method as follows:
ObjectId("507f191e810c19729de860ea").toString()
This operation will return the following output:

ObjectId("507f191e810c19729de860ea")
To return the value of an ObjectId() object as a hexadecimal string, use the valueOf() method as follows:
ObjectId("507f191e810c19729de860ea").valueOf()
22.2. ObjectId
195
This operation returns the following output:

507f191e810c19729de860ea
22.3 GridFS
GridFS is a specication for storing and retrieving les that exceed the BSON -document size limit (page 1105) of 16MB. Instead of storing a le in a single document, GridFS divides a le into parts, or chunks, 5 and stores each of those chunks as a separate document. By default GridFS limits chunk size to 256k. GridFS uses two collections to store les. One collection stores the le chunks, and the other stores le metadata. When you query a GridFS store for a le, the driver or client will reassemble the chunks as needed. You can perform range queries on les stored through GridFS. You also can access information from arbitrary sections of les, which allows you to skip into the middle of a video or audio le. GridFS is useful not only for storing les that exceed 16MB but also for storing any les for which you want access without having to load the entire le into memory. For more information on the indications of GridFS, see When should I use GridFS? (page 716).
22.3.1 Implement GridFS

To store and retrieve les using GridFS, use either of the following: A MongoDB driver. See the drivers (page 555) documentation for information on using GridFS with your driver. The mongofiles (page 1075) command-line tool in the mongo (page 1036) shell. (page 1075). See mongoles
22.3.2 GridFS Collections

GridFS stores les in two collections: chunks stores the binary chunks. For details, see The chunks Collection (page 1117). files stores the les metadata. For details, see The les Collection (page 1118). GridFS places the collections in a common bucket by prexing each with the bucket name. By default, GridFS uses two collections with names prexed by fs bucket: fs.files fs.chunks You can choose a different bucket name than fs, and create multiple buckets in a single database. Each document in the chunks collection represents a distinct chunk of a le as represented in the GridFS store. Each chunk is identied by its unique ObjectID stored in its _id eld. For descriptions of all elds in the chunks and files collections, see GridFS Reference (page 1117).
5
The use of the term chunks in the context of GridFS is not related to the use of the term chunks in the context of sharding.
196
22.3.3 GridFS Index

GridFS uses a unique, compound index on the chunks collection for the files_id and n elds. The files_id eld contains the _id of the chunks parent document. The n eld contains the sequence number of the chunk. GridFS numbers all chunks, starting with 0. For descriptions of the documents and elds in the chunks collection, see GridFS Reference (page 1117). The GridFS index allows efcient retrieval of chunks using the files_id and n values, as shown in the following example:
cursor = db.fs.chunks.find({files_id: myFileID}).sort({n:1});
See the relevant driver (page 555) documentation for the specic behavior of your GridFS application. If your driver does not create this index, issue the following operation using the mongo (page 1036) shell:
db.fs.chunks.ensureIndex( { files_id: 1, n: 1 }, { unique: true } );
22.3.4 Example Interface

The following is an example of the GridFS interface in Java. The example is for demonstration purposes only. For API specics, see the relevant driver (page 555) documentation. By default, the interface must support the default GridFS bucket, named fs, as in the following:
// returns default GridFS bucket (i.e. "fs" collection) GridFS myFS = new GridFS(myDatabase); // saves the file to "fs" GridFS bucket myFS.createFile(new File("/tmp/largething.mpg"));
Optionally, interfaces may support other additional GridFS buckets as in the following example:
// returns GridFS bucket named "contracts" GridFS myContracts = new GridFS(myDatabase, "contracts"); // retrieve GridFS object "smithco" GridFSDBFile file = myContracts.findOne("smithco"); // saves the GridFS file to the file system file.writeTo(new File("/tmp/smithco.pdf"));
22.4 Database References

MongoDB does not support joins. In MongoDB some data is denormalized, or stored with related data in documents to remove the need for joins. However, in some cases it makes sense to store related information in separate documents, typically in different collections or databases. MongoDB applications use one of two methods for relating documents: 1. Manual references (page 1115) where you save the _id eld of one document in another document as a reference. Then your application can run a second query to return the embedded data. These references are simple and sufcient for most use cases. 2. DBRefs (page 1116) are references from one document to another using the value of the rst documents _id eld collection, and optional database name. To resolve DBRefs, your application must perform additional
22.4. Database References
197
queries to return the referenced documents. Many drivers (page 555) have helper methods that form the query for the DBRef automatically. The drivers 6 do not automatically resolve DBRefs into documents. Use a DBRef when you need to embed documents from multiple collections in documents from one collection. DBRefs also provide a common format and type to represent these relationships among documents. The DBRef format provides common semantics for representing links between documents if your database must interact with multiple frameworks and tools. Unless you have a compelling reason for using a DBRef, use manual references.
22.4.1 Manual References

Background Manual references refers to the practice of including one documents _id eld in another document. The application can then issue a second query to resolve the referenced elds as needed. Process Consider the following operation to insert two documents, using the _id eld of the rst document as a reference in the second document:
original_id = ObjectId() db.places.insert({ "_id": original_id "name": "Broadway Center" "url": "bc.example.net" }) db.people.insert({ "name": "Erin" "places_id": original_id "url": "bc.example.net/Erin" })
Then, when a query returns the document from the people collection you can, if needed, make a second query for the document referenced by the places_id eld in the places collection. Use For nearly every case where you want to store a relationship between two documents, use manual references (page 1115). The references are simple to create and your application can resolve references as needed. The only limitation of manual linking is that these references do not convey the database and collection name. If you have documents in a single collection that relate to documents in more than one collection, you may need to consider using DBRefs (page 1116).
6
Some community supported drivers may have alternate behavior and may resolve a DBRef into a document automatically.
198
22.4.2 DBRefs
Background DBRefs are a convention for representing a document, rather than a specic reference type. They include the name of the collection, and in some cases the database, in addition to the value from the _id eld. Format DBRefs have the following elds: $ref The $ref eld holds the name of the collection where the referenced document resides. $id The $id eld contains the value of the _id eld in the referenced document. $db Optional. Contains the name of the database where the referenced document resides. Only some drivers support $db references. Example DBRef document would resemble the following:
{ "$ref" : <value>, "$id" : <value>, "$db" : <value> }
Consider a document from a collection that stored a DBRef in a creator eld:

{ "_id" : ObjectId("5126bbf64aed4daf9e2ab771"), // .. application fields "creator" : { "$ref" : "creators", "$id" : ObjectId("5126bc054aed4daf9e2ab772"), "$db" : "users" } }
The DBRef in this example, points to a document in the creators collection of the users database that has ObjectId("5126bc054aed4daf9e2ab772") in its _id eld. Note: The order of elds in the DBRef matters, and you must use the above sequence when using a DBRef.
Support C++ The C++ driver contains no support for DBRefs. You can transverse references manually. C# The C# driver provides access to DBRef objects with the MongoDBRef Class and supplies the FetchDBRef Method for accessing these objects. Java The DBRef class provides supports for DBRefs from Java. JavaScript The mongo (page 1036) shells JavaScript (page 917) interface provides a DBRef.
199
Perl The Perl driver contains no support for DBRefs. You can transverse references manually or use the MongoDBx::AutoDeref CPAN module. PHP The PHP driver does support DBRefs, including the optional $db reference, through The MongoDBRef class. Python The Python driver provides the DBRef class, and the dereference method for interacting with DBRefs. Ruby The Ruby Driver supports DBRefs using the DBRef class and the deference method. Use In most cases you should use the manual reference (page 1115) method for connecting two or more related documents. However, if you need to reference documents from multiple collections, consider a DBRef.
200
CHAPTER 23
CRUD Operations for MongoDB
These documents provide an overview and examples of common database operations, i.e. CRUD, in MongoDB.
23.1 Create
Of the four basic database operations (i.e. CRUD), create operations are those that add new records or documents to a collection in MongoDB. For general information about write operations and the factors that affect their performance, see Write Operations (page 179); for documentation of the other CRUD operations, see the Core MongoDB Operations (CRUD) (page 165) page. Overview (page 201) insert() (page 202) Insert the First Document in a Collection (page 202) Insert a Document without Specifying an _id Field (page 203) Bulk Insert Multiple Documents (page 205) Insert a Document with save() (page 206) update() Operations with the upsert Flag (page 207) Insert a Document that Contains field and value Pairs (page 207) Insert a Document that Contains Update Operator Expressions (page 208) Update operations with save() (page 209)
23.1.1 Overview
You can create documents in a MongoDB collection using any of the following basic operations: insert (page 202) updates with the upsert option (page 207) All insert operations in MongoDB exhibit the following properties: If you attempt to insert a document without the _id eld, the client library or the mongod (page 1021) instance will add an _id eld and populate the eld with a unique ObjectId. For operations with write concern (page 398), if you specify an _id eld, the _id eld must be unique within the collection; otherwise the mongod (page 1021) will return a duplicate key exception. The maximum BSON document size is 16 megabytes.
201
The maximum document size helps ensure that a single document cannot use excessive amount of RAM or, during transmission, excessive amount of bandwidth. To store documents larger than the maximum size, MongoDB provides the GridFS API. See mongofiles (page 1075) and the documentation for your driver (page 555) for more information about GridFS. Documents (page 187) have the following restrictions on eld names: The eld name _id is reserved for use as a primary key; its value must be unique in the collection, is immutable, and may be of any type other than an array. The eld names cannot start with the $ character. The eld names cannot contain the . character. Note: As of these driver versions (page 1185), all write operations will issue a getLastError (page 831) command to conrm the result of the write operation:
{ getLastError: 1 }
Refer to the documentation on write concern (page 180) in the Write Operations (page 179) document for more information.
23.1.2 insert()
The insert() (page 936) is the primary method to insert a document or documents into a MongoDB collection, and has the following syntax:
db.collection.insert( <document> )
Corresponding Operation in SQL The insert() (page 936) method is analogous to the INSERT statement.
Insert the First Document in a Collection If the collection does not exist 1 , then the insert() (page 936) method creates the collection during the rst insert. Specically in the example, if the collection bios does not exist , then the insert operation will create this collection:
db.bios.insert( { _id: 1, name: { first: John, last: Backus }, birth: new Date(Dec 03, 1924), death: new Date(Mar 17, 2007), contribs: [ Fortran, ALGOL, Backus-Naur Form, FP ], awards: [ { award: W.W. McDowell Award, year: 1967, by: IEEE Computer Society }, { award: National Medal of Science, year: 1975,
1
You can also view a list of the existing collections in the database using the show collections operation in the mongo (page 1036) shell.
202
Chapter 23. CRUD Operations for MongoDB
by: National Science Foundation }, { award: Turing Award, year: 1977, by: ACM }, { award: Draper Prize, year: 1993, by: National Academy of Engineering } ] } )
You can conrm the insert by querying (page 209) the bios collection:
db.bios.find()
This operation returns the following document from the bios collection:
{ "_id" : 1, "name" : { "first" : "John", "last" : "Backus" }, "birth" : ISODate("1924-12-03T05:00:00Z"), "death" : ISODate("2007-03-17T04:00:00Z"), "contribs" : [ "Fortran", "ALGOL", "Backus-Naur Form", "FP" ], "awards" : [ { "award" : "W.W. McDowell Award", "year" : 1967, "by" : "IEEE Computer Society" }, { "award" : "National Medal of Science", "year" : 1975, "by" : "National Science Foundation" }, { "award" : "Turing Award", "year" : 1977, "by" : "ACM" }, { "award" : "Draper Prize", "year" : 1993, "by" : "National Academy of Engineering" } ] }
Insert a Document without Specifying an _id Field If the new document does not contain an _id eld, then the insert() (page 936) method adds the _id eld to the document and generates a unique ObjectId for the value:
23.1. Create
203
db.bios.insert( { name: { first: John, last: McCarthy }, birth: new Date(Sep 04, 1927), death: new Date(Dec 24, 2011), contribs: [ Lisp, Artificial Intelligence, ALGOL ], awards: [ { award: Turing Award, year: 1971, by: ACM }, { award: Kyoto Prize, year: 1988, by: Inamori Foundation }, { award: National Medal of Science, year: 1990, by: National Science Foundation } ] } )
You can verify the inserted document by the querying the bios collection:
db.bios.find( { name: { first: John, last: McCarthy } } )
The returned document contains an _id eld with the generated ObjectId value:
{ "_id" : ObjectId("50a1880488d113a4ae94a94a"), "name" : { "first" : "John", "last" : "McCarthy" }, "birth" : ISODate("1927-09-04T04:00:00Z"), "death" : ISODate("2011-12-24T05:00:00Z"), "contribs" : [ "Lisp", "Artificial Intelligence", "ALGOL" ], "awards" : [ { "award" : "Turing Award", "year" : 1971, "by" : "ACM" }, { "award" : "Kyoto Prize", "year" :1988, "by" : "Inamori Foundation" }, { "award" : "National Medal of Science", "year" : 1990, "by" : "National Science Foundation" } ] }
204
Bulk Insert Multiple Documents If you pass an array of documents to the insert() (page 936) method, the insert() (page 936) performs a bulk insert into a collection. The following operation inserts three documents into the bios collection. The operation also illustrates the dynamic schema characteristic of MongoDB. Although the document with _id: 3 contains a eld title which does not appear in the other documents, MongoDB does not require the other documents to contain this eld:
db.bios.insert( [ { _id: 3, name: { first: Grace, last: Hopper }, title: Rear Admiral, birth: new Date(Dec 09, 1906), death: new Date(Jan 01, 1992), contribs: [ UNIVAC, compiler, FLOW-MATIC, COBOL ], awards: [ { award: Computer Sciences Man of the Year, year: 1969, by: Data Processing Management Association }, { award: Distinguished Fellow, year: 1973, by: British Computer Society }, { award: W. W. McDowell Award, year: 1976, by: IEEE Computer Society }, { award: National Medal of Technology, year: 1991, by: United States } ] }, { _id: 4, name: { first: Kristen, last: Nygaard }, birth: new Date(Aug 27, 1926), death: new Date(Aug 10, 2002), contribs: [ OOP, Simula ], awards: [ { award: Rosing Prize, year: 1999, by: Norwegian Data Association }, { award: Turing Award, year: 2001, by: ACM }, {
23.1. Create
205
award: IEEE John von Neumann Medal, year: 2001, by: IEEE } ] }, { _id: 5, name: { first: Ole-Johan, last: Dahl }, birth: new Date(Oct 12, 1931), death: new Date(Jun 29, 2002), contribs: [ OOP, Simula ], awards: [ { award: Rosing Prize, year: 1999, by: Norwegian Data Association }, { award: Turing Award, year: 2001, by: ACM }, { award: IEEE John von Neumann Medal, year: 2001, by: IEEE } ] } ] )
Insert a Document with save() The save() (page 946) method performs an insert if the document to save does not contain the _id eld. The following save() (page 946) operation performs an insert into the bios collection since the document does not contain the _id eld:
db.bios.save( { name: { first: Guido, last: van Rossum}, birth: new Date(Jan 31, 1956), contribs: [ Python ], awards: [ { award: Award for the Advancement of Free Software, year: 2001, by: Free Software Foundation }, { award: NLUUG Award, year: 2003, by: NLUUG } ]
206
} )
23.1.3 update() Operations with the upsert Flag

The update() (page 948) operation in MongoDB accepts an upsert ag that modies the behavior of update() (page 948) from updating existing documents (page 219), to inserting data. These update() (page 948) operations with the upsert ag eliminate the need to perform an additional operation to check for existence of a record before performing either an update or an insert operation. These update operations have the use <query> argument to determine the write operation: If the query matches an existing document(s), the operation is an update (page 219). If the query matches no document in the collection, the operation is an insert (page 201). An upsert operation has the following syntax 2 :
db.collection.update( <query>, <update>, { upsert: true } )
Insert a Document that Contains field and value Pairs If no document matches the <query> argument, the upsert performs an insert. If the <update> argument includes only eld and value pairs, the new document contains the elds and values specied in the <update> argument. If query does not include an _id eld, the operation adds the _id eld and generates a unique ObjectId for its value. The following update inserts a new document into the bios collection 2 :
db.bios.update( { name: { first: Dennis, last: Ritchie} }, { name: { first: Dennis, last: Ritchie}, birth: new Date(Sep 09, 1941), death: new Date(Oct 12, 2011), contribs: [ UNIX, C ], awards: [ { award: Turing Award, year: 1983, by: ACM }, { award: National Medal of Technology, year: 1998, by: United States }, { award: Japan Prize, year: 2011, by: The Japan Prize Foundation }
2 Prior to version 2.2, in the mongo (page 1036) shell, you would specify the upsert and the multi options in the update() (page 948) method as positional boolean options. See update() (page 948) for details.
23.1. Create
207
] }, { upsert: true } )
Insert a Document that Contains Update Operator Expressions If no document matches the <query> argument, the update operation inserts a new document. If the <update> argument includes only update operators (page 784), the new document contains the elds and values from <query> argument with the operations from the <update> argument applied. The following operation inserts a new document into the bios collection 2 :
db.bios.update( { _id: 7, name: { first: Ken, last: Thompson } }, { $set: { birth: new Date(Feb 04, 1943), contribs: [ UNIX, C, B, UTF-8 ], awards: [ { award: Turing Award, year: 1983, by: ACM }, { award: IEEE Richard W. Hamming Medal, year: 1990, by: IEEE }, { award: National Medal of Technology, year: 1998, by: United States }, { award: Tsutomu Kanai Award, year: 1999, by: IEEE }, { award: Japan Prize, year: 2011, by: The Japan Prize Foundation } ] } }, { upsert: true } )
208
Update operations with save() The save() (page 946) method is identical to an update operation with the upsert ag (page 207) performs an upsert if the document to save contains the _id eld. To determine whether to perform an insert or an update, save() (page 946) method queries documents on the _id eld. The following operation performs an upsert that inserts a document into the bios collection since no documents in the collection contains an _id eld with the value 10:
db.bios.save( { _id: 10, name: { first: Yukihiro, aka: Matz, last: Matsumoto}, birth: new Date(Apr 14, 1965), contribs: [ Ruby ], awards: [ { award: Award for the Advancement of Free Software, year: 2011, by: Free Software Foundation } ] } )
23.2 Read
Of the four basic database operations (i.e. CRUD), read operations are those that retrieve records or documents from a collection in MongoDB. For general information about read operations and the factors that affect their performance, see Read Operations (page 167); for documentation of the other CRUD operations, see the Core MongoDB Operations (CRUD) (page 165) page.
23.2. Read
209
Overview (page 210) find() (page 210) Return All Documents in a Collection (page 211) Return Documents that Match Query Conditions (page 212) * Equality Matches (page 212) * Using Operators (page 212) * Query for Ranges (page 212) * On Arrays (page 213) Query an Element (page 213) Query Multiple Fields on an Array of Documents (page 213) On Subdocuments (page 213) * Exact Matches (page 213) Fields of a Subdocument (page 214) Logical Operators (page 214) * OR Disjunctions (page 214) AND Conjunctions (page 214) With a Projection (page 215) * Specify the Fields to Return (page 215) * Explicitly Exclude the _id Field (page 215) * Return All but the Excluded Fields (page 215) * On Arrays and Subdocuments (page 215) Iterate the Returned Cursor (page 216) * With Variable Name (page 216) * With next() Method (page 216) * With forEach() Method (page 217) Modify the Cursor Behavior (page 217) * Order Documents in the Result Set (page 217) * Limit the Number of Documents to Return (page 217) * Set the Starting Point of the Result Set (page 218) * Combine Cursor Methods (page 218) findOne() (page 218) With Empty Query Specication (page 218) With a Query Specication (page 218) With a Projection (page 219) * Specify the Fields to Return (page 219) * Return All but the Excluded Fields (page 219) Access the findOne Result (page 219)
23.2.1 Overview
You can retrieve documents from MongoDB using either of the following methods: nd (page 210) ndOne (page 218)
23.2.2 find()
The find() (page 924) method is the primary method to select documents from a collection. The find() (page 924) method returns a cursor that contains a number of documents. Most drivers (page 555) provide application developers with a native iterable interface for handling cursors and accessing documents. The find() (page 924) method has the following syntax:
210
db.collection.find( <query>, <projection> )
Corresponding Operation in SQL The find() (page 924) method is analogous to the SELECT statement, while: the <query> argument corresponds to the WHERE statement, and the <projection> argument corresponds to the list of elds to select from the result set. The examples refer to a collection named bios that contains documents with the following prototype:
{ "_id" : 1, "name" : { "first" : "John", "last" :"Backus" }, "birth" : ISODate("1924-12-03T05:00:00Z"), "death" : ISODate("2007-03-17T04:00:00Z"), "contribs" : [ "Fortran", "ALGOL", "Backus-Naur Form", "FP" ], "awards" : [ { "award" : "W.W. McDowellAward", "year" : 1967, "by" : "IEEE Computer Society" }, { "award" : "National Medal of Science", "year" : 1975, "by" : "National Science Foundation" }, { "award" : "Turing Award", "year" : 1977, "by" : "ACM" }, { "award" : "Draper Prize", "year" : 1993, "by" : "National Academy of Engineering" } ] }
Note: In the mongo (page 1036) shell, you can format the output by adding .pretty() to the find() (page 924) method call.
Return All Documents in a Collection If there is no <query> argument, the find() (page 924) method selects all documents from a collection. The following operation returns all documents (or more precisely, a cursor to all documents) in the bios collection:
db.bios.find()
23.2. Read
211
Return Documents that Match Query Conditions If there is a <query> argument, the find() (page 924) method selects all documents from a collection that satises the query specication.
Equality Matches
The following operation returns a cursor to documents in the bios collection where the eld _id equals 5:
db.bios.find( { _id: 5 } )
Using Operators
The following operation returns a cursor to all documents in the bios collection where the eld _id equals 5 or ObjectId("507c35dd8fada716c89d0013"):
db.bios.find( { _id: { $in: [ 5, } )
ObjectId("507c35dd8fada716c89d0013") ] }
Query for Ranges
You may combine comparison operators to specify ranges:

db.collection.find( { field: { $gt: value1, $lt: value2 } } );
This statement returns all documents with field between value1 and value2. Note: If the eld contains an array and the query has multiple conditional operators, the eld as a whole will match if either a single array element meets the conditions or a combination of array elements meet the conditions. Example Query a eld that contains an array. A collection students contains the following documents where the score eld contains an array of values:
{ "_id" : 1, "score" : [ -1, 3 ] } { "_id" : 2, "score" : [ 1, 5 ] } { "_id" : 3, "score" : [ 5, 5 ] }
Then, the following query:

db.students.find( { score: { $gt: 0, $lt: 2 } } )
Will match the following documents:

{ "_id" : 1, "score" : [ -1, 3 ] } { "_id" : 2, "score" : [ 1, 5 ] }
212
In the document with _id equal to 1, the score: [ -1, 3 ] as a whole meets the specied conditions since the element -1 meets the $lt: 2 condition and the element 3 meets the $gt: 0 condition. In the document with _id equal to 2, the score: [ 1, 5 ] as a whole meets the specied conditions since the element 1 meets both the $lt: 2 condition and the $gt: 0 condition.
On Arrays
Query an Element The following operation returns a cursor to all documents in the bios collection where the array eld contribs contains the element UNIX:
db.bios.find( { contribs: UNIX } )
Query Multiple Fields on an Array of Documents The following operation returns a cursor to all documents in the bios collection where awards array contains a subdocument element that contains the award eld equal to Turing Award and the year eld greater than 1980:
db.bios.find( { awards: { $elemMatch: { award: Turing Award, year: { $gt: 1980 } } } } )
On Subdocuments
Exact Matches The following operation returns a cursor to all documents in the bios collection where the subdocument name is exactly { first: Yukihiro, last: Matsumoto }, including the order:
db.bios.find( { name: { first: Yukihiro, last: Matsumoto } } )
The name eld must match the sub-document exactly, including order. For instance, the query would not match documents with name elds that held either of the following values:
{ first: Yukihiro, aka: Matz, last: Matsumoto }
23.2. Read
213
{ last: Matsumoto, first: Yukihiro }
Fields of a Subdocument The following operation returns a cursor to all documents in the bios collection where the subdocument name contains a eld first with the value Yukihiro and a eld last with the value Matsumoto; the query uses dot notation to access elds in a subdocument:
db.bios.find( { name.first: Yukihiro, name.last: Matsumoto } )
The query matches the document where the name eld contains a subdocument with the eld first with the value Yukihiro and a eld last with the value Matsumoto. For instance, the query would match documents with name elds that held either of the following values:
{ first: Yukihiro, aka: Matz, last: Matsumoto } { last: Matsumoto, first: Yukihiro }
Logical Operators
OR Disjunctions The following operation returns a cursor to all documents in the bios collection where either the eld first in the sub-document name starts with the letter G or where the eld birth is less than new Date(01/01/1945):
db.bios.find( { $or: [ { name.first : /^G/ }, { birth: { $lt: new Date(01/01/1945) } } ] } )
AND Conjunctions The following operation returns a cursor to all documents in the bios collection where the eld first in the subdocument name starts with the letter K and the array eld contribs contains the element UNIX:
db.bios.find( { name.first: /^K/, contribs: UNIX } )
214
In this query, the parameters (i.e. the selections of both elds) combine using an implicit logical AND for criteria on different elds contribs and name.first. For multiple AND criteria on the same eld, use the $and (page 768) operator. With a Projection If there is a <projection> argument, the find() (page 924) method returns only those elds as specied in the <projection> argument to include or exclude: Note: The _id eld is implicitly included in the <projection> argument. In projections that explicitly include elds, _id is the only eld that you can explicitly exclude. Otherwise, you cannot mix include eld and exclude eld specications.
Specify the Fields to Return
The following operation nds all documents in the bios collection and returns only the name eld, the contribs eld, and the _id eld:
db.bios.find( { }, { name: 1, contribs: 1 } )
Explicitly Exclude the _id Field
The following operation nds all documents in the bios collection and returns only the name eld and the contribs eld:
db.bios.find( { }, { name: 1, contribs: 1, _id: 0 } )
Return All but the Excluded Fields
The following operation nds the documents in the bios collection where the contribs eld contains the element OOP and returns all elds except the _id eld, the first eld in the name subdocument, and the birth eld from the matching documents:
db.bios.find( { contribs: OOP }, { _id: 0, name.first: 0, birth: 0 } )
On Arrays and Subdocuments
The following operation nds all documents in the bios collection and returns the last eld in the name subdocument and the rst two elements in the contribs array:
23.2. Read
215
db.bios.find( { }, { _id: 0, name.last: 1, contribs: { $slice: 2 } } )
See also: dot notation for information on reaching into embedded sub-documents. Arrays (page 170) for more examples on accessing arrays. Subdocuments (page 169) for more examples on accessing subdocuments. $elemMatch (page 783) query operator for more information on matching array elements. $elemMatch (page 798) projection operator for additional information on restricting array elements to return. Iterate the Returned Cursor The find() (page 924) method returns a cursor to the results; however, in the mongo (page 1036) shell, if the returned cursor is not assigned to a variable, then the cursor is automatically iterated up to 20 times 3 to print up to the rst 20 documents that match the query, as in the following example:
db.bios.find( { _id: 1 } );
With Variable Name
When you assign the find() (page 924) to a variable, you can type the name of the cursor variable to iterate up to 20 times 1 and print the matching documents, as in the following example:
var myCursor = db.bios.find( { _id: 1 } ); myCursor
With next() Method
You can use the cursor method next() (page 962) to access the documents, as in the following example:
var myCursor = db.bios.find( { _id: 1 } ); var myDocument = myCursor.hasNext() ? myCursor.next() : null; if (myDocument) { var myName = myDocument.name; print (tojson(myName)); }
To print, you can also use the printjson() method instead of print(tojson()):
3 You can use the DBQuery.shellBatchSize to change the number of iteration from the default value 20. See Cursor Flags (page 177) and Cursor Behaviors (page 176) for more information.
216
if (myDocument) { var myName = myDocument.name; printjson(myName); }
With forEach() Method
You can use the cursor method forEach() (page 958) to iterate the cursor and access the documents, as in the following example:
var myCursor = db.bios.find( { _id: 1 } ); myCursor.forEach(printjson);
For more information on cursor handling, see: cursor.hasNext() (page 958) cursor.next() (page 962) cursor.forEach() (page 958) cursors (page 175) JavaScript cursor methods (page 951) Modify the Cursor Behavior In addition to the <query> and the <projection> arguments, the mongo (page 1036) shell and the drivers (page 555) provide several cursor methods that you can call on the cursor returned by find() (page 924) method to modify its behavior, such as:
Order Documents in the Result Set
The sort() (page 965) method orders the documents in the result set. The following operation returns all documents (or more precisely, a cursor to all documents) in the bios collection ordered by the name eld ascending:
db.bios.find().sort( { name: 1 } )
sort() (page 965) corresponds to the ORDER BY statement in SQL.

Limit the Number of Documents to Return
The limit() (page 959) method limits the number of documents in the result set. The following operation returns at most 5 documents (or more precisely, a cursor to at most 5 documents) in the bios collection:
db.bios.find().limit( 5 )
limit() (page 959) corresponds to the LIMIT statement in SQL.
23.2. Read
217
Set the Starting Point of the Result Set
The skip() (page 964) method controls the starting point of the results set. The following operation returns all documents, skipping the rst 5 documents in the bios collection:
db.bios.find().skip( 5 )
Combine Cursor Methods
You can chain these cursor methods, as in the following examples 4 :

db.bios.find().sort( { name: 1 } ).limit( 5 ) db.bios.find().limit( 5 ).sort( { name: 1 } )
See the JavaScript cursor methods (page 951) reference and your driver (page 555) documentation for additional references. See Cursors (page 175) for more information regarding cursors.
23.2.3 findOne()
The findOne() (page 929) method selects a single document from a collection and returns that document. findOne() (page 929) does not return a cursor. The findOne() (page 929) method has the following syntax:
db.collection.findOne( <query>, <projection> )
Except for the return value, findOne() (page 929) method is quite similar to the find() (page 924) method; in fact, internally, the findOne() (page 929) method is the find() (page 924) method with a limit of 1. With Empty Query Specication If there is no <query> argument, the findOne() (page 929) method selects just one document from a collection. The following operation returns a single document from the bios collection:
db.bios.findOne()
With a Query Specication If there is a <query> argument, the findOne() (page 929) method selects the rst document from a collection that meets the <query> argument: The following operation returns the rst matching document from the bios collection where either the eld first in the subdocument name starts with the letter G or where the eld birth is less than new Date(01/01/1945):
db.bios.findOne( { $or: [ { name.first : /^G/ }, { birth: { $lt: new Date(01/01/1945) } }
Regardless of the order you chain the limit() (page 959) and the sort() (page 965), the request to the server has the structure that treats the query and the sort() (page 965) modier as a single object. Therefore, the limit() (page 959) operation method is always applied after the sort() (page 965) regardless of the specied order of the operations in the chain. See the meta query operators (page 801) for more information.
4
218
] } )
With a Projection You can pass a <projection> argument to findOne() (page 929) to control the elds included in the result set.
Specify the Fields to Return
The following operation nds a document in the bios collection and returns only the name eld, the contribs eld, and the _id eld:
db.bios.findOne( { }, { name: 1, contribs: 1 } )
Return All but the Excluded Fields
The following operation returns a document in the bios collection where the contribs eld contains the element OOP and returns all elds except the _id eld, the first eld in the name subdocument, and the birth eld from the matching documents:
db.bios.findOne( { contribs: OOP }, { _id: 0, name.first: 0, birth: 0 } )
Access the findOne Result Although similar to the find() (page 924) method, because the findOne() (page 929) method returns a document rather than a cursor, you cannot apply the cursor methods such as limit() (page 959), sort() (page 965), and skip() (page 964) to the result of the findOne() (page 929) method. However, you can access the document directly, as in the example:
var myDocument = db.bios.findOne(); if (myDocument) { var myName = myDocument.name; print (tojson(myName)); }
23.3 Update
Of the four basic database operations (i.e. CRUD), update operations are those that modify existing records or documents in a MongoDB collection. For general information about write operations and the factors that affect their performance, see Write Operations (page 179); for documentation of other CRUD operations, see the Core MongoDB Operations (CRUD) (page 165) page. 23.3. Update 219
Overview (page 220) Update (page 220) Modify with Update Operators (page 221) * Update a Field in a Document (page 221) * Add a New Field to a Document (page 221) * Remove a Field from a Document (page 222) * Update Arrays (page 222) Update an Element by Specifying Its Position (page 222) Update an Element without Specifying Its Position (page 222) Update a Document Element without Specifying Its Position (page 222) Add an Element to an Array (page 223) Update Multiple Documents (page 223) * Replace Existing Document with New Document (page 223) update() Operations with the upsert Flag (page 224) Save (page 224) Behavior (page 225) Save Performs an Update (page 225) Update Operators (page 226) Fields (page 226) Array (page 226) * Operators (page 226) * Modiers (page 226) Bitwise (page 226) Isolation (page 227)
23.3.1 Overview
Update operation modies an existing document or documents in a collection. MongoDB provides the following methods to perform update operations: update (page 220) save (page 224) Note: Consider the following behaviors of MongoDBs update operations. When performing update operations that increase the document size beyond the allocated space for that document, the update operation relocates the document on disk and may reorder the document elds depending on the type of update. As of these driver versions (page 1185), all write operations will issue a getLastError (page 831) command to conrm the result of the write operation:
{ getLastError: 1 }
Refer to the documentation on write concern (page 180) in the Write Operations (page 179) document for more information.
23.3.2 Update
The update() (page 948) method is the primary method used to modify documents in a MongoDB collection. By default, the update() (page 948) method updates a single document, but by using the multi option, update()
220
(page 948) can update all documents that match the query criteria in the collection. The update() (page 948) method can either replace the existing document with the new document or update specic elds in the existing document. The update() (page 948) has the following syntax 5 :
db.collection.update( <query>, <update>, <options> )
Corresponding operation in SQL The update() (page 948) method corresponds to the UPDATE operation in SQL, and: the <query> argument corresponds to the WHERE statement, and the <update> corresponds to the SET ... statement. The default behavior of the update() (page 948) method updates a single document and would correspond to the SQL UPDATE statement with the LIMIT 1. With the multi option, update() (page 948) method would correspond to the SQL UPDATE statement without the LIMIT clause.
Modify with Update Operators If the <update> argument contains only update operator (page 226) expressions such as the $set (page 788) operator expression, the update() (page 948) method updates the corresponding elds in the document. To update elds in subdocuments, MongoDB uses dot notation.
Update a Field in a Document
Use $set (page 788) to update a value of a eld. The following operation queries the bios collection for the rst document that has an _id eld equal to 1 and sets the value of the eld middle, in the subdocument name, to Warner:
db.bios.update( { _id: 1 }, { $set: { name.middle: Warner }, } )
Add a New Field to a Document
If the <update> argument contains elds not currently in the document, the update() (page 948) method adds the new elds to the document. The following operation queries the bios collection for the rst document that has an _id eld equal to 3 and adds to that document a new mbranch eld and a new aka eld in the subdocument name:
db.bios.update( { _id: 3 }, { $set: { mbranch: Navy, name.aka: Amazing Grace
5 This examples uses the interface added in MongoDB 2.2 to specify the multi and the upsert options in a document form. Prior to version 2.2, in the mongo (page 1036) shell, you would specify the upsert and the multi options in the update() (page 948) method as positional boolean options. See update() (page 948) for details.
23.3. Update
221
} } )
Remove a Field from a Document
If the <update> argument contains $unset (page 788) operator, the update() (page 948) method removes the eld from the document. The following operation queries the bios collection for the rst document that has an _id eld equal to 3 and removes the birth eld from the document:
db.bios.update( { _id: 3 }, { $unset: { birth: 1 } } )
Update Arrays
Update an Element by Specifying Its Position If the update operation requires an update of an element in an array eld, the update() (page 948) method can perform the update using the position of the element and dot notation. Arrays in MongoDB are zero-based. The following operation queries the bios collection for the rst document with _id eld equal to 1 and updates the second element in the contribs array:
db.bios.update( { _id: 1 }, { $set: { contribs.1: ALGOL 58 } } )
Update an Element without Specifying Its Position The update() (page 948) method can perform the update using the $ (page 788) positional operator if the position is not known. The array eld must appear in the query argument in order to determine which array element to update. The following operation queries the bios collection for the rst document where the _id eld equals 3 and the contribs array contains an element equal to compiler. If found, the update() (page 948) method updates the rst matching element in the array to A compiler in the document:
db.bios.update( { _id: 3, contribs: compiler }, { $set: { contribs.$: A compiler } } )
Update a Document Element without Specifying Its Position The update() (page 948) method can perform the update of an array that contains subdocuments by using the positional operator (i.e. $ (page 788)) and the dot notation. The following operation queries the bios collection for the rst document where the _id eld equals 6 and the awards array contains a subdocument element with the by eld equal to ACM. If found, the update() (page 948) method updates the by eld in the rst matching subdocument:
222
db.bios.update( { _id: 6, awards.by: ACM } , { $set: { awards.$.by: Association for Computing Machinery } } )
Add an Element to an Array The following operation queries the bios collection for the rst document that has an _id eld equal to 1 and adds a new element to the awards eld:
db.bios.update( { _id: 1 }, { $push: { awards: { award: IBM Fellow, year: 1963, by: IBM } } } )
Update Multiple Documents
If the <options> argument contains the multi option set to true or 1, the update() (page 948) method updates all documents that match the query. The following operation queries the bios collection for all documents where the awards eld contains a subdocument element with the award eld equal to Turing and sets the turing eld to true in the matching documents 6 :
db.bios.update( { awards.award: Turing }, { $set: { turing: true } }, { multi: true } )
Replace Existing Document with New Document If the <update> argument contains only eld and value pairs, the update() (page 948) method replaces the existing document with the document in the <update> argument, except for the _id eld. The following operation queries the bios collection for the rst document that has a name eld equal to { first: John, last: McCarthy } and replaces all but the _id eld in the document with the elds in the <update> argument:
db.bios.update( { name: { first: John, last: McCarthy } }, { name: { first: Ken, last: Iverson }, born: new Date(Dec 17, 1941), died: new Date(Oct 19, 2004), contribs: [ APL, J ], awards: [ { award: Turing Award, year: 1979, by: ACM }, { award: Harry H. Goode Memorial Award, year: 1975, by: IEEE Computer Society },
6 Prior to version 2.2, in the mongo (page 1036) shell, you would specify the upsert and the multi options in the update() (page 948) method as positional boolean options. See update() (page 948) for details.
23.3. Update
223
{ award: IBM Fellow, year: 1970, by: IBM } ] } )
23.3.3 update() Operations with the upsert Flag

If you set the upsert option in the <options> argument to true or 1 and no existing document match the <query> argument, the update() (page 948) method can insert a new document into the collection. 7 The following operation queries the bios collection for a document with the _id eld equal to 11 and the name eld equal to { first: James, last: Gosling}. If the query selects a document, the operation performs an update operation. If a document is not found, update() (page 948) inserts a new document containing the elds and values from <query> argument with the operations from the <update> argument applied. 8
db.bios.update( { _id:11, name: { first: James, last: Gosling } }, { $set: { born: new Date(May 19, 1955), contribs: [ Java ], awards: [ { award: The Economist Innovation Award, year: 2002, by: The Economist }, { award: Officer of the Order of Canada, year: 2007, by: Canada } ] } }, { upsert: true } )
See also Update Operations with the Upsert Flag (page 207) in the Create (page 201) document.
23.3.4 Save
The save() (page 946) method performs a special type of update() (page 948), depending on the _id eld of the specied document. The save() (page 946) method has the following syntax:
db.collection.save( <document> )
7 Prior to version 2.2, in the mongo (page 1036) shell, you would specify the upsert and the multi options in the update() (page 948) method as positional boolean options. See update() (page 948) for details. 8 If the <update> argument includes only eld and value pairs, the new document contains the elds and values specied in the <update> argument. If the <update> argument includes only update operators (page 226), the new document contains the elds and values from <query> argument with the operations from the <update> argument applied.
224
Behavior If you specify a document with an _id eld, save() (page 946) performs an update() (page 948) with the upsert option set: if an existing document in the collection has the same _id, save() (page 946) updates that document, and inserts the document otherwise. If you do not specify a document with an _id eld to save() (page 946), performs an insert() (page 936) operation. That is, save() (page 946) method is equivalent to the update() (page 948) method with the upsert option and a <query> argument with an _id eld. Example Consider the following pseudocode explanation of save() (page 946) as an illustration of its behavior:
function save( doc ) { if( doc["_id"] ) { update( {_id: doc["_id"] }, doc, { upsert: true } ); } else { insert(doc); } }
Save Performs an Update If the <document> argument contains the _id eld that exists in the collection, the save() (page 946) method performs an update that replaces the existing document with the <document> argument. The following operation queries the bios collection for a document where the _id equals ObjectId("507c4e138fada716c89d0014") and replaces the document with the <document> argument:
db.bios.save( { _id: ObjectId("507c4e138fada716c89d0014"), name: { first: Martin, last: Odersky }, contribs: [ Scala ] } )
See also: Insert a Document with save() (page 206) and Update operations with save() (page 209) in the Create (page 201) section.
23.3. Update
225
23.3.5 Update Operators

Fields Name $inc (page 784) $rename (page 784) $setOnInsert (page 787) $set (page 788) $unset (page 788) Array
Operators
Description Increments the value of the eld by the specied amount. Renames a eld. Sets the value of a eld upon documentation creation during an upsert. Has no effect on update operations that modify existing documents. Sets the value of a eld in an existing document. Removes the specied eld from an existing document.
Name $ (page 788) $addToSet (page 789) $pop (page 790) $pullAll (page 790) $pull (page 791) $pushAll (page 791) $push (page 792)
Modiers
Description Acts as a placeholder to update the rst element that matches the query condition in an update. Adds elements to an existing array only if they do not already exist in the set. Removes the rst or last item of an array. Removes multiple values from an array. Removes items from an array that match a query statement. Deprecated. Adds several items to an array. Adds an item to an array.
Name $each (page 793) $slice (page 793) $sort (page 794) Bitwise Name $bit (page 795)
Description Modies the $push (page 792) and $addToSet (page 789) operators to append multiple items for array updates. Modies the $push (page 792) operator to limit the size of updated arrays. Modies the $push (page 792) operator to reorder documents stored in an array.
Description Performs bitwise AND and OR updates of integer values.
226
Isolation Name $isolated (page 796) Description Modies behavior of multi-updates to improve the isolation of the operation.
23.4 Delete
Of the four basic database operations (i.e. CRUD), delete operations are those that remove documents from a collection in MongoDB. For general information about write operations and the factors that affect their performance, see Write Operations (page 179); for documentation of other CRUD operations, see the Core MongoDB Operations (CRUD) (page 165) page. Overview (page 227) Remove All Documents that Match a Condition (page 228) Remove a Single Document that Matches a Condition (page 228) Remove All Documents from a Collection (page 228) Capped Collection (page 228) Isolation (page 228)
23.4.1 Overview
The remove() (page 227) method in the mongo (page 1036) shell provides this operation, as do corresponding methods in the drivers (page 555). Note: As of these driver versions (page 1185), all write operations will issue a getLastError (page 831) command to conrm the result of the write operation:
{ getLastError: 1 }
Refer to the documentation on write concern (page 180) in the Write Operations (page 179) document for more information. Use the remove() (page 944) method to delete documents from a collection. The remove() (page 944) method has the following syntax:
db.collection.remove( <query>, <justOne> )
Corresponding operation in SQL The remove() (page 944) method is analogous to the DELETE statement, and: the <query> argument corresponds to the WHERE statement, and the <justOne> argument takes a Boolean and has the same affect as LIMIT 1. remove() (page 944) deletes documents from the collection. If you do not specify a query, remove() (page 944) removes all documents from a collection, but does not remove the indexes. 9
9 To remove all documents from a collection, it may be more efcient to use the drop() (page 921) method to drop the entire collection, including the indexes, and then recreate the collection and rebuild the indexes.
23.4. Delete
227
Note: For large deletion operations, it may be more efcient to copy the documents that you want to keep to a new collection and then use drop() (page 921) on the original collection.
23.4.2 Remove All Documents that Match a Condition

If there is a <query> argument, the remove() (page 944) method deletes from the collection all documents that match the argument. The following operation deletes all documents from the bios collection where the subdocument name contains a eld first whose value starts with G:
db.bios.remove( { name.first : /^G/ } )
23.4.3 Remove a Single Document that Matches a Condition

If there is a <query> argument and you specify the <justOne> argument as true or 1, remove() (page 944) only deletes a single document from the collection that matches the query. The following operation deletes a single document from the bios collection where the turing eld equals true:
db.bios.remove( { turing: true }, 1 )
23.4.4 Remove All Documents from a Collection

If there is no <query> argument, the remove() (page 944) method deletes all documents from a collection. The following operation deletes all documents from the bios collection:
db.bios.remove()
Note: This operation is not equivalent to the drop() (page 921) method.
23.4.5 Capped Collection

You cannot use the remove() (page 944) method with a capped collection.
23.4.6 Isolation
If the <query> argument to the remove() (page 944) method matches multiple documents in the collection, the delete operation may interleave with other write operations to that collection. For an unsharded collection, you have the option to override this behavior with the $isolated (page 796) isolation operator, effectively isolating the delete operation from other write operations. To isolate the operation, include $isolated: 1 in the <query> parameter as in the following example:
db.bios.remove( { turing: true, $isolated: 1 } )
228
Part V
Data Modeling
229
Data in MongoDB has a exible schema. Collections do not enforce document structure. Although you may be able to use different structures for a single data set in MongoDB, different data models may have signicant impacts on MongoDB and application performance. Consider Data Modeling Considerations for MongoDB Applications (page 233) for a conceptual overview of data modeling problems in MongoDB, and the Data Modeling Patterns (page 239) documents for examples of different approaches to data models. See also: Use Cases (page 607) for overviews of application design, including data models, with MongoDB.
231
232
CHAPTER 24
Background
24.1 Data Modeling Considerations for MongoDB Applications

24.1.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. This means that: documents in the same collection do not need to have the same set of elds or structure, and common elds in a collections documents may hold different types of data. Each document only needs to contain relevant elds to the entity or object that the document represents. In practice, most documents in a collection share a similar structure. Schema exibility means that you can model your documents in MongoDB so that they can closely resemble and reect application-level objects. As in all data modeling, when developing data models (i.e. schema designs) for MongoDB, you must consider the inherent properties and requirements of the application objects and the relationships between application objects. MongoDB data models must also reect: how data will grow and change over time, and the kinds of queries your application will perform. These considerations and requirements force developers to make a number of multi-factored decisions when modeling data, including: normalization and de-normalization. These decisions reect the degree to which the data model should store related pieces of data in a single document. Fully normalized data models describe relationships using references (page 1115) between documents, while de-normalized models may store redundant information across related models. indexing strategy (page 341). representation of data in arrays in BSON . Although a number of data models may be functionally equivalent for a given application, different data models may have signicant impacts on MongoDB and applications performance. This document provides a high level overview of these data modeling decisions and factors. In addition, consider the Data Modeling Patterns and Examples (page 237) section which provides more concrete examples of all the discussed patterns.
233
24.1.2 Data Modeling Decisions

Data modeling decisions involve determining how to structure the documents to model the data effectively. The primary decision is whether to embed (page 234) or to use references (page 234). Embedding To de-normalize data, store two related pieces of data in a single document. Operations within a document are less expensive for the server than operations that involve multiple documents. In general, use embedded data models when: you have contains relationships between entities. See Model Embedded One-to-One Relationships Between Documents (page 239). you have one-to-many relationships where the many objects always appear with or are viewed in the context of their parent documents. See Model Embedded One-to-Many Relationships Between Documents (page 240). Embedding provides the following benets: generally better performance for read operations. the ability to request and retrieve related data in a single database operation. Embedding related data in documents, can lead to situations where documents grow after creation. Document growth can impact write performance and lead to data fragmentation. Furthermore, documents in MongoDB must be smaller than the maximum BSON document size (page 1105). For larger documents, consider using GridFS (page 196). See also: dot notation for information on reaching into embedded sub-documents. Arrays (page 170) for more examples on accessing arrays. Subdocuments (page 169) for more examples on accessing subdocuments. Referencing To normalize data, store references (page 1115) between two documents to indicate a relationship between the data represented in each document. In general, use normalized data models: when embedding would result in duplication of data but would not provide sufcient read performance advantages to outweigh the implications of the duplication. to represent more complex many-to-many relationships. to model large hierarchical data sets. See data-modeling-trees. Referencing provides more exibility than embedding; however, to resolve the references, client-side applications must issue follow-up queries. In other words, using references requires more roundtrips to the server. See Model Referenced One-to-Many Relationships Between Documents (page 241) for an example of referencing.
234
Chapter 24. Background
Atomicity MongoDB only provides atomic operations on the level of a single document. 1 As a result needs for atomic operations inuence decisions to use embedded or referenced relationships when modeling data for MongoDB. Embed elds that need to be modied together atomically in the same document. See Model Data for Atomic Operations (page 243) for an example of atomic updates within a single document.
24.1.3 Operational Considerations

In addition to normalization and normalization concerns, a number of other operational factors help shape data modeling decisions in MongoDB. These factors include: data lifecycle management, number of collections and indexing requirements, sharding, and managing document growth. These factors implications for database and application performance as well as future maintenance and development costs. Data Lifecycle Management Data modeling decisions should also take data lifecycle management into consideration. The Time to Live or TTL feature (page 577) of collections expires documents after a period of time. Consider using the TTL feature if your application requires some data to persist in the database for a limited period of time. Additionally, if your application only uses recently inserted documents consider Capped Collections (page 558). Capped collections provide rst-in-rst-out (FIFO) management of inserted documents and optimized to support operations that insert and read documents based on insertion order. Large Number of Collections In certain situations, you might choose to store information in several collections rather than in a single collection. Consider a sample collection logs that stores log documents for various environment and applications. The logs collection contains documents of the following form:
{ log: "dev", ts: ..., info: ... } { log: "debug", ts: ..., info: ...}
If the total number of documents is low you may group documents into collection by type. For logs, consider maintaining distinct log collections, such as logs.dev and logs.debug. The logs.dev collection would contain only the documents related to the dev environment. Generally, having large number of collections has no signicant performance penalty and results in very good performance. Distinct collections are very important for high-throughput batch processing. When using models that have a large number of collections, consider the following behaviors: Each collection has a certain minimum overhead of a few kilobytes.
Document-level atomic operations include all operations within a single MongoDB document record: operations that affect multiple subdocuments within that single record are still atomic.
1
24.1. Data Modeling Considerations for MongoDB Applications
235
Each index, including the index on _id, requires at least 8KB of data space. A single <database>.ns le stores all meta-data for each database. Each index and collection has its own entry in the namespace le, MongoDB places limits on the size of namespace files (page 1105). Because of limits on namespaces (page 1105), you may wish to know the current number of namespaces in order to determine how many additional namespaces the database can support, as in the following example:
db.system.namespaces.count()
The <database>.ns le defaults to 16 MB. To change the size of the <database>.ns le, pass a new size to --nssize option <new size MB> (page 1026) on server start. The --nssize (page 1026) sets the size for new <database>.ns les. For existing databases, after starting up the server with --nssize (page 1026), run the db.repairDatabase() (page 984) command from the mongo (page 1036) shell. Indexes Create indexes to support common queries. Generally, indexes and index use in MongoDB correspond to indexes and index use in relational database: build indexes on elds that appear often in queries and for all operations that return sorted results. MongoDB automatically creates a unique index on the _id eld. As you create indexes, consider the following behaviors of indexes: Each index requires at least 8KB of data space. Adding an index has some negative performance impact for write operations. For collections with high writeto-read ratio, indexes are expensive as each insert must add keys to each index. Collections with high proportion of read operations to write operations often benet from additional indexes. Indexes do not affect un-indexed read operations. See Indexing Strategies (page 341) for more information on determining indexes. Additionally, the MongoDB database proler (page 99) may help identify inefcient queries. Sharding Sharding allows users to partition a collection within a database to distribute the collections documents across a number of mongod (page 1021) instances or shards. The shard key determines how MongoDB distributes data among shards in a sharded collection. Selecting the proper shard key (page 485) has signicant implications for performance. See Sharded Cluster Overview (page 485) for more information on sharding and the selection of the shard key (page 485). Document Growth Certain updates to documents can increase the document size, such as pushing elements to an array and adding new elds. If the document size exceeds the allocated space for that document, MongoDB relocates the document on disk. This internal relocation can be both time and resource consuming. Although MongoDB automatically provides padding to minimize the occurrence of relocations, you may still need to manually handle document growth. Refer to Pre-Aggregated Reports (page 619) for an example of the Pre-allocation approach to handle document growth.
236
24.1.4 Data Modeling Patterns and Examples

The following documents provide overviews of various data modeling patterns and common schema design considerations: Model Embedded One-to-One Relationships Between Documents (page 239) Model Embedded One-to-Many Relationships Between Documents (page 240) Model Referenced One-to-Many Relationships Between Documents (page 241) Model Data for Atomic Operations (page 243) Model Tree Structures with Parent References (page 244) Model Tree Structures with Child References (page 244) Model Tree Structures with Materialized Paths (page 246) Model Tree Structures with Nested Sets (page 247) For more information and examples of real-world data modeling, consider the following external resources: Schema Design by Example Walkthrough MongoDB Data Modeling Document Design for MongoDB Dynamic Schema Blog Post MongoDB Data Modeling and Rails Ruby Example of Materialized Paths Sean Cribs Blog Post which was the source for much of the data-modeling-trees content.
24.1. Data Modeling Considerations for MongoDB Applications
237
238
CHAPTER 25
Data Modeling Patterns
25.1 Model Embedded One-to-One Relationships Between Documents

25.1.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. Decisions that affect how you model data can affect application performance and database capacity. See Data Modeling Considerations for MongoDB Applications (page 233) for a full high level overview of data modeling in MongoDB. This document describes a data model that uses embedded (page 234) documents to describe relationships between connected data.
25.1.2 Pattern
Consider the following example that maps patron and address relationships. The example illustrates the advantage of embedding over referencing if you need to view one data entity in context of the other. In this one-to-one relationship between patron and address data, the address belongs to the patron. In the normalized data model, the address contains a reference to the parent.
{ _id: "joe", name: "Joe Bookreader" } { patron_id: "joe", street: "123 Fake Street", city: "Faketon", state: "MA" zip: 12345 }
If the address data is frequently retrieved with the name information, then with referencing, your application needs to issue multiple queries to resolve the reference. The better data model would be to embed the address data in the patron data, as in the following document:
{ _id: "joe", name: "Joe Bookreader",
239
address: { street: "123 Fake Street", city: "Faketon", state: "MA" zip: 12345 } }
With the embedded data model, your application can retrieve the complete patron information with one query.
25.2 Model Embedded One-to-Many Relationships Between Documents

25.2.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. Decisions that affect how you model data can affect application performance and database capacity. See Data Modeling Considerations for MongoDB Applications (page 233) for a full high level overview of data modeling in MongoDB. This document describes a data model that uses embedded (page 234) documents to describe relationships between connected data.
25.2.2 Pattern
Consider the following example that maps patron and multiple address relationships. The example illustrates the advantage of embedding over referencing if you need to view many data entities in context of another. In this one-tomany relationship between patron and address data, the patron has multiple address entities. In the normalized data model, the address contains a reference to the parent.
{ _id: "joe", name: "Joe Bookreader" } { patron_id: "joe", street: "123 Fake Street", city: "Faketon", state: "MA", zip: 12345 } { patron_id: "joe", street: "1 Some Other Street", city: "Boston", state: "MA", zip: 12345 }
If your application frequently retrieves the address data with the name information, then your application needs to issue multiple queries to resolve the references. A more optimal schema would be to embed the address data entities in the patron data, as in the following document:
240
Chapter 25. Data Modeling Patterns
{ _id: "joe", name: "Joe Bookreader", addresses: [ { street: "123 Fake Street", city: "Faketon", state: "MA", zip: 12345 }, { street: "1 Some Other Street", city: "Boston", state: "MA", zip: 12345 } ] }
With the embedded data model, your application can retrieve the complete patron information with one query.
25.3 Model Referenced One-to-Many Relationships Between Documents

25.3.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. Decisions that affect how you model data can affect application performance and database capacity. See Data Modeling Considerations for MongoDB Applications (page 233) for a full high level overview of data modeling in MongoDB. This document describes a data model that uses references (page 234) between documents to describe relationships between connected data.
25.3.2 Pattern
Consider the following example that maps publisher and book relationships. The example illustrates the advantage of referencing over embedding to avoid repetition of the publisher information. Embedding the publisher document inside the book document would lead to repetition of the publisher data, as the following documents show:
{ title: "MongoDB: The Definitive Guide", author: [ "Kristina Chodorow", "Mike Dirolf" ], published_date: ISODate("2010-09-24"), pages: 216, language: "English", publisher: { name: "OReilly Media", founded: 1980, location: "CA" } }
25.3. Model Referenced One-to-Many Relationships Between Documents
241
{ title: "50 Tips and Tricks for MongoDB Developer", author: "Kristina Chodorow", published_date: ISODate("2011-05-06"), pages: 68, language: "English", publisher: { name: "OReilly Media", founded: 1980, location: "CA" } }
To avoid repetition of the publisher data, use references and keep the publisher information in a separate collection from the book collection. When using references, the growth of the relationships determine where to store the reference. If the number of books per publisher is small with limited growth, storing the book reference inside the publisher document may sometimes be useful. Otherwise, if the number of books per publisher is unbounded, this data model would lead to mutable, growing arrays, as in the following example:
{ name: "OReilly Media", founded: 1980, location: "CA", books: [12346789, 234567890, ...] } { _id: 123456789, title: "MongoDB: The Definitive Guide", author: [ "Kristina Chodorow", "Mike Dirolf" ], published_date: ISODate("2010-09-24"), pages: 216, language: "English" } { _id: 234567890, title: "50 Tips and Tricks for MongoDB Developer", author: "Kristina Chodorow", published_date: ISODate("2011-05-06"), pages: 68, language: "English" }
To avoid mutable, growing arrays, store the publisher reference inside the book document:
{ _id: "oreilly", name: "OReilly Media", founded: 1980, location: "CA" } { _id: 123456789, title: "MongoDB: The Definitive Guide", author: [ "Kristina Chodorow", "Mike Dirolf" ],
242
published_date: ISODate("2010-09-24"), pages: 216, language: "English", publisher_id: "oreilly" } { _id: 234567890, title: "50 Tips and Tricks for MongoDB Developer", author: "Kristina Chodorow", published_date: ISODate("2011-05-06"), pages: 68, language: "English", publisher_id: "oreilly" }
25.4 Model Data for Atomic Operations

25.4.1 Pattern
Consider the following example that keeps a library book and its checkout information. The example illustrates how embedding elds related to an atomic update within the same document ensures that the elds are in sync. Consider the following book document that stores the number of available copies for checkout and the current checkout information:
book = { _id: 123456789, title: "MongoDB: The Definitive Guide", author: [ "Kristina Chodorow", "Mike Dirolf" ], published_date: ISODate("2010-09-24"), pages: 216, language: "English", publisher_id: "oreilly", available: 3, checkout: [ { by: "joe", date: ISODate("2012-10-15") } ] }
You can use the db.collection.findAndModify() (page 926) method to atomically determine if a book is available for checkout and update with the new checkout information. Embedding the available eld and the checkout eld within the same document ensures that the updates to these elds are in sync:
db.books.findAndModify ( { query: { _id: 123456789, available: { $gt: 0 } }, update: { $inc: { available: -1 }, $push: { checkout: { by: "abc", date: new Date() } } } } )
25.4. Model Data for Atomic Operations
243
25.5 Model Tree Structures with Parent References

25.5.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. Decisions that affect how you model data can affect application performance and database capacity. See Data Modeling Considerations for MongoDB Applications (page 233) for a full high level overview of data modeling in MongoDB. This document describes a data model that describes a tree-like structure in MongoDB documents by storing references (page 234) to parent nodes in children nodes.
25.5.2 Pattern
The Parent References pattern stores each tree node in a document; in addition to the tree node, the document stores the id of the nodes parent. Consider the following example that models a tree of categories using Parent References:
db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( { { { { { { _id: _id: _id: _id: _id: _id: "MongoDB", parent: "Databases" } ) "Postgres", parent: "Databases" } ) "Databases", parent: "Programming" } ) "Languages", parent: "Programming" } ) "Programming", parent: "Books" } ) "Books", parent: null } )
The query to retrieve the parent of a node is fast and straightforward:

db.categories.findOne( { _id: "MongoDB" } ).parent
You can create an index on the eld parent to enable fast search by the parent node:
db.categories.ensureIndex( { parent: 1 } )
You can query by the parent eld to nd its immediate children nodes:
db.categories.find( { parent: "Databases" } )
The Parent Links pattern provides a simple solution to tree storage, but requires multiple queries to retrieve subtrees.
25.6 Model Tree Structures with Child References

25.6.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. Decisions that affect how you model data can affect application performance and database capacity. See Data Modeling Considerations for MongoDB Applications (page 233) for a full high level overview of data modeling in MongoDB. This document describes a data model that describes a tree-like structure in MongoDB documents by storing references (page 234) in the parent-nodes to children nodes.
25.6.2 Pattern
The Child References pattern stores each tree node in a document; in addition to the tree node, document stores in an array the id(s) of the nodes children. 244 Chapter 25. Data Modeling Patterns
Consider the following example that models a tree of categories using Child References:
db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( { { { { { { _id: _id: _id: _id: _id: _id: "MongoDB", children: [] } ) "Postgres", children: [] } ) "Databases", children: [ "MongoDB", "Postgres" ] } ) "Languages", children: [] } ) "Programming", children: [ "Databases", "Languages" ] } ) "Books", children: [ "Programming" ] } )
The query to retrieve the immediate children of a node is fast and straightforward:
db.categories.findOne( { _id: "Databases" } ).children
You can create an index on the eld children to enable fast search by the child nodes:
db.categories.ensureIndex( { children: 1 } )
You can query for a node in the children eld to nd its parent node as well as its siblings:
db.categories.find( { children: "MongoDB" } )
The Child References pattern provides a suitable solution to tree storage as long as no operations on subtrees are necessary. This pattern may also provide a suitable solution for storing graphs where a node may have multiple parents.
25.7 Model Tree Structures with an Array of Ancestors

25.7.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. Decisions that affect how you model data can affect application performance and database capacity. See Data Modeling Considerations for MongoDB Applications (page 233) for a full high level overview of data modeling in MongoDB. This document describes a data model that describes a tree-like structure in MongoDB documents using references (page 234) to parent nodes and an array that stores all ancestors.
25.7.2 Pattern
The Array of Ancestors pattern stores each tree node in a document; in addition to the tree node, document stores in an array the id(s) of the nodes ancestors or path. Consider the following example that models a tree of categories using Array of Ancestors:
db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( { { { { { { _id: _id: _id: _id: _id: _id:
"MongoDB", ancestors: [ "Books", "Programming", "Databases" ], parent: " "Postgres", ancestors: [ "Books", "Programming", "Databases" ], parent: "Databases", ancestors: [ "Books", "Programming" ], parent: "Programming "Languages", ancestors: [ "Books", "Programming" ], parent: "Programming "Programming", ancestors: [ "Books" ], parent: "Books" } ) "Books", ancestors: [ ], parent: null } )
The query to retrieve the ancestors or path of a node is fast and straightforward:
db.categories.findOne( { _id: "MongoDB" } ).ancestors
You can create an index on the eld ancestors to enable fast search by the ancestors nodes:
25.7. Model Tree Structures with an Array of Ancestors
245
db.categories.ensureIndex( { ancestors: 1 } )
You can query by the ancestors to nd all its descendants:

db.categories.find( { ancestors: "Programming" } )
The Array of Ancestors pattern provides a fast and efcient solution to nd the descendants and the ancestors of a node by creating an index on the elements of the ancestors eld. This makes Array of Ancestors a good choice for working with subtrees. The Array of Ancestors pattern is slightly slower than the Materialized Paths pattern but is more straightforward to use.
25.8 Model Tree Structures with Materialized Paths

25.8.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. Decisions that affect how you model data can affect application performance and database capacity. See Data Modeling Considerations for MongoDB Applications (page 233) for a full high level overview of data modeling in MongoDB. This document describes a data model that describes a tree-like structure in MongoDB documents by storing full relationship paths between documents.
25.8.2 Pattern
The Materialized Paths pattern stores each tree node in a document; in addition to the tree node, document stores as a string the id(s) of the nodes ancestors or path. Although the Materialized Paths pattern requires additional steps of working with strings and regular expressions, the pattern also provides more exibility in working with the path, such as nding nodes by partial paths. Consider the following example that models a tree of categories using Materialized Paths ; the path string uses the comma , as a delimiter:
db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( { { { { { { _id: _id: _id: _id: _id: _id: "Books", path: null } ) "Programming", path: ",Books," } ) "Databases", path: ",Books,Programming," } ) "Languages", path: ",Books,Programming," } ) "MongoDB", path: ",Books,Programming,Databases," } ) "Postgres", path: ",Books,Programming,Databases," } )
You can query to retrieve the whole tree, sorting by the path:
db.categories.find().sort( { path: 1 } )
You can use regular expressions on the path eld to nd the descendants of Programming:
db.categories.find( { path: /,Programming,/ } )
You can also retrieve the descendants of Books where the Books is also at the topmost level of the hierarchy:
db.categories.find( { path: /^,Books,/ } )
To create an index on the eld path use the following invocation:
246
db.categories.ensureIndex( { path: 1 } )
This index may improve performance, depending on the query: For queries of the Books sub-tree (e.g. http://docs.mongodb.org/manual/^,Books,/) an index on the path eld improves the query performance signicantly. For queries of the Programming sub-tree (e.g. http://docs.mongodb.org/manual/,Programming,/), or similar queries of sub-tress, where the node might be in the middle of the indexed string, the query must inspect the entire index. For these queries an index may provide some performance improvement if the index is signicantly smaller than the entire collection.
25.9 Model Tree Structures with Nested Sets

25.9.1 Overview
Data in MongoDB has a exible schema. Collections do not enforce document structure. Decisions that affect how you model data can affect application performance and database capacity. See Data Modeling Considerations for MongoDB Applications (page 233) for a full high level overview of data modeling in MongoDB. This document describes a data model that describes a tree like structure that optimizes discovering subtrees at the expense of tree mutability.
25.9.2 Pattern
The Nested Sets pattern identies each node in the tree as stops in a round-trip traversal of the tree. The application visits each node in the tree twice; rst during the initial trip, and second during the return trip. The Nested Sets pattern stores each tree node in a document; in addition to the tree node, document stores the id of nodes parent, the nodes initial stop in the left eld, and its return stop in the right eld. Consider the following example that models a tree of categories using Nested Sets:
db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( db.categories.insert( { { { { { { _id: _id: _id: _id: _id: _id: "Books", parent: 0, left: 1, right: 12 } ) "Programming", parent: "Books", left: 2, right: 11 } ) "Languages", parent: "Programming", left: 3, right: 4 } ) "Databases", parent: "Programming", left: 5, right: 10 } ) "MongoDB", parent: "Databases", left: 6, right: 7 } ) "Postgres", parent: "Databases", left: 8, right: 9 } )
You can query to retrieve the descendants of a node:

var databaseCategory = db.v.findOne( { _id: "Databases" } ); db.categories.find( { left: { $gt: databaseCategory.left }, right: { $lt: databaseCategory.right } }
The Nested Sets pattern provides a fast and efcient solution for nding subtrees but is inefcient for modifying the tree structure. As such, this pattern is best for static trees that do not change.
25.10 Model Data to Support Keyword Search

Note: Keyword search is not the same as text search or full text search, and does not provide stemming or other text-processing features. See the Limitations of Keyword Indexes (page 248) section for more information.
25.9. Model Tree Structures with Nested Sets
247
In 2.4, MongoDB provides a text search feature. See Text Search (page 371) for more information. If your application needs to perform queries on the content of a eld that holds text you can perform exact matches on the text or use $regex (page 774) to use regular expression pattern matches. However, for many operations on text, these methods do not satisfy application requirements. This pattern describes one method for supporting keyword search using MongoDB to support application search functionality, that uses keywords stored in an array in the same document as the text eld. Combined with a multi-key index (page 332), this pattern can support applications keyword search operations.
25.10.1 Pattern
To add structures to your document to support keyword-based queries, create an array eld in your documents and add the keywords as strings in the array. You can then create a multi-key index (page 332) on the array and create queries that select values from the array. Example Given a collection of library volumes that you want to provide topic-based search. For each volume, you add the array topics, and you add as many keywords as needed for a given volume. For the Moby-Dick volume you might have the following document:
{ title : "Moby-Dick" , author : "Herman Melville" , published : 1851 , ISBN : 0451526996 , topics : [ "whaling" , "allegory" , "revenge" , "American" , "novel" , "nautical" , "voyage" , "Cape Cod" ] }
You then create a multi-key index on the topics array:

db.volumes.ensureIndex( { topics: 1 } )
The multi-key index creates separate index entries for each keyword in the topics array. For example the index contains one entry for whaling and another for allegory. You then query based on the keywords. For example:
db.volumes.findOne( { topics : "voyage" }, { title: 1 } )
Note: An array with a large number of elements, such as one with several hundreds or thousands of keywords will incur greater indexing costs on insertion.
25.10.2 Limitations of Keyword Indexes

MongoDB can support keyword searches using specic data models and multi-key indexes (page 332); however, these keyword indexes are not sufcient or comparable to full-text products in the following respects: Stemming. Keyword queries in MongoDB can not parse keywords for root or related words. Synonyms. Keyword-based search features must provide support for synonym or related queries in the application layer. Ranking. The keyword look ups described in this document do not provide a way to weight results. 248 Chapter 25. Data Modeling Patterns
Asynchronous Indexing. MongoDB builds indexes synchronously, which means that the indexes used for keyword indexes are always current and can operate in real-time. However, asynchronous bulk indexes may be more efcient for some kinds of content and workloads.
25.10. Model Data to Support Keyword Search
249
250
Part VI
Aggregation
251
In version 2.2, MongoDB introduced the aggregation framework (page 255) that provides a powerful and exible set of tools to use for many data aggregation tasks. If youre familiar with data aggregation in SQL, consider the SQL to Aggregation Framework Mapping Chart (page 307) document as an introduction to some of the basic concepts in the aggregation framework. Consider the full documentation of the aggregation framework and other data aggregation tools for MongoDB here:
253
254
CHAPTER 26
Aggregation Framework
New in version 2.1.
26.1 Overview
The MongoDB aggregation framework provides a means to calculate aggregated values without having to use mapreduce. While map-reduce is powerful, it is often more difcult than necessary for many simple aggregation tasks, such as totaling or averaging eld values. If youre familiar with SQL, the aggregation framework provides similar functionality to GROUP BY and related SQL operators as well as simple forms of self joins. Additionally, the aggregation framework provides projection capabilities to reshape the returned data. Using the projections in the aggregation framework, you can add computed elds, create new virtual sub-objects, and extract sub-elds into the top-level of results. See also: Consider Aggregation Framework Examples (page 261) and Aggregation Framework Reference (page 271) for more documentation.
26.2 Framework Components

This section provides an introduction to the two concepts that underpin the aggregation framework: pipelines and expressions.
26.2.1 Pipelines
Conceptually, documents from a collection pass through an aggregation pipeline, which transforms these objects as they pass through. For those familiar with UNIX-like shells (e.g. bash,) the concept is analogous to the pipe (i.e. |) used to string text lters together. In a shell environment the pipe redirects a stream of characters from the output of one process to the input of the next. The MongoDB aggregation pipeline streams MongoDB documents from one pipeline operator (page 291) to the next to process the documents. Pipeline operators can be repeated in the pipe. All pipeline operators process a stream of documents and the pipeline behaves as if the operation scans a collection and passes all matching documents into the top of the pipeline. Each operator in the pipeline transforms each document as it passes through the pipeline. Note: Pipeline operators need not produce one output document for every input document: operators may also 255
generate new documents or lter out documents. Warning: The pipeline cannot operate on values of the following types: Binary, Symbol, MinKey, MaxKey, DBRef, Code, and CodeWScope. See also: The Aggregation Framework Reference (page 271) includes documentation of the following pipeline operators: $project (page 285) $match (page 281) $limit (page 280) $skip (page 287) $unwind (page 290) $group (page 278) $sort (page 287) $geoNear (page 276)
26.2.2 Expressions
Expressions (page 300) produce output documents based on calculations performed on input documents. The aggregation framework denes expressions using a document format using prexes. Expressions are stateless and are only evaluated when seen by the aggregation process. All aggregation expressions can only operate on the current document in the pipeline, and cannot integrate data from other documents. The accumulator expressions used in the $group (page 278) operator maintain that state (e.g. totals, maximums, minimums, and related data) as documents progress through the pipeline. See also: Aggregation expressions (page 300) for additional examples of the expressions provided by the aggregation framework.
26.3 Use
26.3.1 Invocation
Invoke an aggregation operation with the aggregate() (page 918) wrapper in the mongo (page 1036) shell or the aggregate (page 808) database command. Always call aggregate() (page 918) on a collection object that determines the input documents of the aggregation pipeline. The arguments to the aggregate() (page 918) method specify a sequence of pipeline operators (page 291), where each operator may have a number of operands. First, consider a collection of documents named articles using the following format:
{ title : "this is my title" , author : "bob" , posted : new Date () , pageViews : 5 , tags : [ "fun" , "good" , "fun" ] , comments : [
256
Chapter 26. Aggregation Framework
{ author :"joe" , text : "this is cool" } , { author :"sam" , text : "this is bad" } ], other : { foo : 5 } }
The following example aggregation operation pivots data to create a set of author names grouped by tags applied to an article. Call the aggregation framework by issuing the following command:
db.articles.aggregate( { $project : { author : 1, tags : 1, } }, { $unwind : "$tags" }, { $group : { _id : { tags : "$tags" }, authors : { $addToSet : "$author" } } } );
The aggregation pipeline begins with the collection articles and selects the author and tags elds using the $project (page 285) aggregation operator. The $unwind (page 290) operator produces one output document per tag. Finally, the $group (page 278) operator pivots these elds.
26.3.2 Result
The aggregation operation in the previous section returns a document with two elds: result which holds an array of documents returned by the pipeline ok which holds the value 1, indicating success. Changed in version 2.4: If an error occurs, the aggregate() (page 918) helper throws an exception. In previous versions, the helper returned a document with the error message and code, and ok status eld not equal to 1, same as the aggregate (page 808) command. As a document, the result is subject to the BSON Document size (page 1105) limit, which is currently 16 megabytes.
26.4 Optimizing Performance

Because you will always call aggregate (page 808) on a collection object, which logically inserts the entire collection into the aggregation pipeline, you may want to optimize the operation by avoiding scanning the entire collection whenever possible.
26.4.1 Pipeline Operators and Indexes

Depending on the order in which they appear in the pipeline, aggregation operators can take advantage of indexes. The following pipeline operators take advantage of an index when they occur at the beginning of the pipeline: $match (page 281) $sort (page 287) $limit (page 280)
26.4. Optimizing Performance
257
$skip (page 287). The above operators can also use an index when placed before the following aggregation operators: $project (page 285) $unwind (page 290) $group (page 278). New in version 2.4. The $geoNear (page 276) pipeline operator takes advantage of a geospatial index. When using $geoNear (page 276), the $geoNear (page 276) pipeline operation must appear as the rst stage in an aggregation pipeline.
26.4.2 Early Filtering

If your aggregation operation requires only a subset of the data in a collection, use the $match (page 281) operator to restrict which items go in to the top of the pipeline, as in a query. When placed early in a pipeline, these $match (page 281) operations use suitable indexes to scan only the matching documents in a collection. Placing a $match (page 281) pipeline stage followed by a $sort (page 287) stage at the start of the pipeline is logically equivalent to a single query with a sort, and can use an index. In future versions there may be an optimization phase in the pipeline that reorders the operations to increase performance without affecting the result. However, at this time place $match (page 281) operators at the beginning of the pipeline when possible.
26.4.3 Pipeline Sequence Optimization

Changed in version 2.4. Aggregation operations have an optimization phase which attempts to re-arrange the pipeline for improved performance. $sort + $skip + $limit Sequence Optimization When you have sequence of $sort (page 287) followed by a $skip (page 287) followed by a $limit (page 280), an optimization occurs whereby the $limit (page 280) moves in front of the $skip (page 287). For example, if the pipeline consists of the following stages:
{ $sort: { age : -1 } }, { $skip: 10 }, { $limit: 5 }
During the optimization phase, the optimizer transforms the sequence to the following:
{ $sort: { age : -1 } }, { $limit: 15 } { $skip: 10 }
Note: The $limit (page 280) value has increased to the sum of the initial value and the $skip (page 287) value.
258
$limit + $skip + $limit + $skip Sequence Optimization When you have continuous sequence of $limit (page 280) pipeline stage followed by a $skip (page 287) pipeline stage, the aggregation will attempt to re-arrange the pipeline stages to combine the limits together and the skips together. For example, if the pipeline consists of the following stages:
{ { { { $limit: 100 }, $skip: 5 }, $limit: 10}, $skip: 2 }
During the intermediate step, the optimizer reverses the position of the $skip (page 287) followed by a $limit (page 280) to $limit (page 280) followed by the $skip (page 287).
{ { { { $limit: 100 }, $limit: 15}, $skip: 5 }, $skip: 2 }
The $limit (page 280) value has increased to the sum of the initial value and the $skip (page 287) value. Then, for the nal $limit (page 280) value, the optimizer selects the minimum between the adjacent $limit (page 280) values. For the nal $skip (page 287) value, the optimizer adds the adjacent $skip (page 287) values, to transform the sequence to the following:
{ $limit: 15 }, { $skip: 7 }
26.4.4 Memory for Cumulative Operators

Certain pipeline operators require access to the entire input set before they can produce any output. For example, $sort (page 287) must receive all of the input from the preceding pipeline operator before it can produce its rst output document. The current implementation of $sort (page 287) does not go to disk in these cases: in order to sort the contents of the pipeline, the entire input must t in memory. Changed in version 2.4: When a $sort (page 287) immediately precedes a $limit (page 280) in the pipeline, the $sort (page 287) operation only maintains the top n results as it progresses, where n is the specied limit. Before 2.4, $sort (page 287) would sort all the results in memory, and then limit the results to n results. $group (page 278) has similar characteristics: Before any $group (page 278) passes its output along the pipeline, it must receive the entirety of its input. For the $group (page 278) operator, this frequently does not require as much memory as $sort (page 287), because it only needs to retain one record for each unique key in the grouping specication. The current implementation of the aggregation framework logs a warning if a cumulative operator consumes 5% or more of the physical memory on the host. Cumulative operators produce an error if they consume 10% or more of the physical memory on the host.
26.5 Sharded Operation

Note: Changed in version 2.1. Some aggregation operations using aggregate (page 808) will cause mongos (page 1032) instances to require more CPU resources than in previous versions. This modied performance prole may dictate alternate architectural decisions if you use the aggregation framework extensively in a sharded environment.
26.5. Sharded Operation
259
The aggregation framework is compatible with sharded collections. When operating on a sharded collection, the aggregation pipeline is split into two parts. The aggregation framework pushes all of the operators up to the rst $group (page 278) or $sort (page 287) operation to each shard. 1 Then, a second pipeline on the mongos (page 1032) runs. This pipeline consists of the rst $group (page 278) or $sort (page 287) and any remaining pipeline operators, and runs on the results received from the shards. The $group (page 278) operator brings in any sub-totals from the shards and combines them: in some cases these may be structures. For example, the $avg (page 272) expression maintains a total and count for each shard; mongos (page 1032) combines these values and then divides.
26.6 Limitations
Aggregation operations with the aggregate (page 808) command have the following limitations: The pipeline cannot operate on values of the following types: Binary, Symbol, MinKey, MaxKey, DBRef, Code, CodeWScope. Output from the pipeline can only contain 16 megabytes. If your result set exceeds this limit, the aggregate (page 808) command produces an error. If any single aggregation operation consumes more than 10 percent of system RAM the operation will produce an error.
1 If an early $match (page 281) can exclude shards through the use of the shard key in the predicate, then these operators are only pushed to the relevant shards.
260
CHAPTER 27
Aggregation Framework Examples
MongoDB provides exible data aggregation functionality with the aggregate (page 808) command. For additional information about aggregation consider the following resources: Aggregation Framework (page 255) Aggregation Framework Reference (page 271) SQL to Aggregation Framework Mapping Chart (page 307) This document provides a number of practical examples that display the capabilities of the aggregation framework. All examples use a publicly available data set of all zipcodes and populations in the United States.
27.1 Requirements
mongod (page 1021) and mongo (page 1036), version 2.2 or later.
27.2 Aggregations using the Zip Code Data Set

To run you will need the zipcode data set. These data are available at: media.mongodb.org/zips.json. mongoimport (page 1056) to load this data set into your mongod (page 1021) instance. Use
27.2.1 Data Model

Each document in this collection has the following form:
{ "_id": "10280", "city": "NEW YORK", "state": "NY", "pop": 5574, "loc": [ -74.016323, 40.710537 ] }
In these documents: The _id eld holds the zipcode as a string.
261
The city eld holds the city. The state eld holds the two letter state abbreviation. The pop eld holds the population. The loc eld holds the location as a latitude longitude pair. All of the following examples use the aggregate() (page 918) helper in the mongo (page 1036) shell. aggregate() (page 918) provides a wrapper around the aggregate (page 808) database command. See the documentation for your driver (page 555) for a more idiomatic interface for data aggregation operations.
27.2.2 States with Populations Over 10 Million

To return all states with a population greater than 10 million, use the following aggregation operation:
db.zipcodes.aggregate( { $group : { _id : "$state", totalPop : { $sum : "$pop" } } }, { $match : {totalPop : { $gte : 10*1000*1000 } } } )
Aggregations operations using the aggregate() (page 918) helper, process all documents on the zipcodes collection. aggregate() (page 918) connects a number of pipeline (page 255) operators, which dene the aggregation process. In the above example, the pipeline passes all documents in the zipcodes collection through the following steps: the $group (page 278) operator collects all documents and creates documents for each state. These new per-state documents have one eld in addition the _id eld: totalPop which is a generated eld using the $sum (page 289) operation to calculate the total value of all pop elds in the source documents. After the $group (page 278) operation the documents in the pipeline resemble the following:
{ "_id" : "AK", "totalPop" : 550043 }
the $match (page 281) operation lters these documents so that the only documents that remain are those where the value of totalPop is greater than or equal to 10 million. The $match (page 281) operation does not alter the documents, which have the same format as the documents output by $group (page 278). The equivalent SQL for this operation is:
SELECT state, SUM(pop) AS pop FROM zips GROUP BY state HAVING pop > (10*1000*1000)
27.2.3 Average City Population by State

To return the average populations for cities in each state, use the following aggregation operation:
db.zipcodes.aggregate( { $group : { _id : { state : "$state", city : "$city" }, pop : { $sum : "$pop" } } }, { $group :
262
Chapter 27. Aggregation Framework Examples
{ _id : "$_id.state", avgCityPop : { $avg : "$pop" } } } )
Aggregations operations using the aggregate() (page 918) helper, process all documents on the zipcodes collection. aggregate() (page 918) a number of pipeline (page 255) operators that dene the aggregation process. In the above example, the pipeline passes all documents in the zipcodes collection through the following steps: the $group (page 278) operator collects all documents and creates new documents for every combination of the city and state elds in the source document. After this stage in the pipeline, the documents resemble the following:
{ "_id" : { "state" : "CO", "city" : "EDGEWATER" }, "pop" : 13154 }
the second $group (page 278) operator collects documents by the state eld and use the $avg (page 272) expression to compute a value for the avgCityPop eld. The nal output of this aggregation operation is:
{ "_id" : "MN", "avgCityPop" : 5335 },
27.2.4 Largest and Smallest Cities by State

To return the smallest and largest cities by population for each state, use the following aggregation operation:
db.zipcodes.aggregate( { $group: { _id: { state: "$state", city: "$city" }, pop: { $sum: "$pop" } } }, { $sort: { pop: 1 } }, { $group: { _id : "$_id.state", biggestCity: { $last: "$_id.city" }, biggestPop: { $last: "$pop" }, smallestCity: { $first: "$_id.city" }, smallestPop: { $first: "$pop" } } }, // the following $project is optional, and // modifies the output format. { $project: { _id: 0, state: "$_id", biggestCity: { name: "$biggestCity", pop: "$biggestPop" }, smallestCity: { name: "$smallestCity", pop: "$smallestPop" } } } )
Aggregations operations using the aggregate() (page 918) helper, process all documents on the zipcodes collection. aggregate() (page 918) a number of pipeline (page 255) operators that dene the aggregation process. All documents from the zipcodes collection pass into the pipeline, which consists of the following steps:
27.2. Aggregations using the Zip Code Data Set
263
the $group (page 278) operator collects all documents and creates new documents for every combination of the city and state elds in the source documents. By specifying the value of _id as a sub-document that contains both elds, the operation preserves the state eld for use later in the pipeline. The documents produced by this stage of the pipeline have a second eld, pop, which uses the $sum (page 289) operator to provide the total of the pop elds in the source document. At this stage in the pipeline, the documents resemble the following:
{ "_id" : { "state" : "CO", "city" : "EDGEWATER" }, "pop" : 13154 }
$sort (page 287) operator orders the documents in the pipeline based on the vale of the pop eld from largest to smallest. This operation does not alter the documents. the second $group (page 278) operator collects the documents in the pipeline by the state eld, which is a eld inside the nested _id document. Within each per-state document this $group (page 278) operator species four elds: Using the $last (page 280) expression, the $group (page 278) operator creates the biggestcity and biggestpop elds that store the city with the largest population and that population. Using the $first (page 276) expression, the $group (page 278) operator creates the smallestcity and smallestpop elds that store the city with the smallest population and that population. The documents, at this stage in the pipeline resemble the following:
{ "_id" : "WA", "biggestCity" : "SEATTLE", "biggestPop" : 520096, "smallestCity" : "BENGE", "smallestPop" : 2 }
The nal operation is $project (page 285), which renames the _id eld to state and moves the biggestCity, biggestPop, smallestCity, and smallestPop into biggestCity and smallestCity sub-documents. The nal output of this aggregation operation is:
{ "state" : "RI", "biggestCity" : { "name" : "CRANSTON", "pop" : 176404 }, "smallestCity" : { "name" : "CLAYVILLE", "pop" : 45 } }
264
27.3 Aggregation with User Preference Data

27.3.1 Data Model
Consider a hypothetical sports club with a database that contains a user collection that tracks users join dates, sport preferences, and stores these data in documents that resemble the following:
{ _id : "jane", joined : ISODate("2011-03-02"), likes : ["golf", "racquetball"] } { _id : "joe", joined : ISODate("2012-07-02"), likes : ["tennis", "golf", "swimming"] }
27.3.2 Normalize and Sort Documents

The following operation returns user names in upper case and in alphabetical order. The aggregation includes user names for all documents in the users collection. You might do this to normalize user names for processing.
db.users.aggregate( [ { $project : { name:{$toUpper:"$_id"} , _id:0 } }, { $sort : { name : 1 } } ] )
All documents from the users collection passes through the pipeline, which consists of the following operations: The $project (page 285) operator: creates a new eld called name. converts the value of the _id to upper case, with the $toUpper (page 290) operator. Then the $project (page 285) creates a new eld, named name to hold this value. suppresses the id eld. $project (page 285) will pass the _id eld by default, unless explicitly suppressed. The $sort (page 287) operator orders the results by the name eld. The results of the aggregation would resemble the following:
{ "name" : "JANE" }, { "name" : "JILL" }, { "name" : "JOE" }
27.3. Aggregation with User Preference Data
265
27.3.3 Return Usernames Ordered by Join Month

The following aggregation operation returns user names sorted by the month they joined. This kind of aggregation could help generate membership renewal notices.
db.users.aggregate( [ { $project : { month_joined : { $month : "$joined" }, name : "$_id", _id : 0 }, { $sort : { month_joined : 1 } } ] )
The pipeline passes all documents in the users collection through the following operations: The $project (page 285) operator: Creates two new elds: month_joined and name. Suppresses the id from the results. The aggregate() (page 918) method includes the _id, unless explicitly suppressed. The $month (page 284) operator converts the values of the joined eld to integer representations of the month. Then the $project (page 285) operator assigns those values to the month_joined eld. The $sort (page 287) operator sorts the results by the month_joined eld. The operation returns results that resemble the following:
{ "month_joined" : 1, "name" : "ruth" }, { "month_joined" : 1, "name" : "harold" }, { "month_joined" : 1, "name" : "kate" } { "month_joined" : 2, "name" : "jill" }
27.3.4 Return Total Number of Joins per Month

The following operation shows how many people joined each month of the year. You might use this aggregated data for such information for recruiting and marketing strategies.
db.users.aggregate( [ { $project : { month_joined : { $month : "$joined" } } } , { $group : { _id : {month_joined:"$month_joined"} , number : { $sum : 1 } } },
266
{ $sort : { "_id.month_joined" : 1 } } ] )
The pipeline passes all documents in the users collection through the following operations: The $project (page 285) operator creates a new eld called month_joined. The $month (page 284) operator converts the values of the joined eld to integer representations of the month. Then the $project (page 285) operator assigns the values to the month_joined eld. The $group (page 278) operator collects all documents with a given month_joined value and counts how many documents there are for that value. Specically, for each unique value, $group (page 278) creates a new per-month document with two elds: _id, which contains a nested document with the month_joined eld and its value. number, which is a generated eld. The $sum (page 289) operator increments this eld by 1 for every document containing the given month_joined value. The $sort (page 287) operator sorts the documents created by $group (page 278) according to the contents of the month_joined eld. The result of this aggregation operation would resemble the following:
{ "_id" : { "month_joined" : 1 }, "number" : 3 }, { "_id" : { "month_joined" : 2 }, "number" : 9 }, { "_id" : { "month_joined" : 3 }, "number" : 5 }
27.3.5 Return the Five Most Common Likes

The following aggregation collects top ve most liked activities in the data set. In this data set, you might use an analysis of this to help inform planning and future development.
db.users.aggregate( [ { $unwind : "$likes" }, { $group : { _id : "$likes" , number : { $sum : 1 } } }, { $sort : { number : -1 } }, { $limit : 5 } ] )
267
The pipeline begins with all documents in the users collection, and passes these documents through the following operations: The $unwind (page 290) operator separates each value in the likes array, and creates a new version of the source document for every element in the array. Example Given the following document from the users collection:
{ _id : "jane", joined : ISODate("2011-03-02"), likes : ["golf", "racquetball"] }
The $unwind (page 290) operator would create the following documents:
{ _id : "jane", joined : ISODate("2011-03-02"), likes : "golf" } { _id : "jane", joined : ISODate("2011-03-02"), likes : "racquetball" }
The $group (page 278) operator collects all documents the same value for the likes eld and counts each grouping. With this information, $group (page 278) creates a new document with two elds: _id, which contains the likes value. number, which is a generated eld. The $sum (page 289) operator increments this eld by 1 for every document containing the given likes value. The $sort (page 287) operator sorts these documents by the number eld in reverse order. The $limit (page 280) operator only includes the rst 5 result documents. The results of aggregation would resemble the following:
{ "_id" : "golf", "number" : 33 }, { "_id" : "racquetball", "number" : 31 }, { "_id" : "swimming", "number" : 24 }, { "_id" : "handball", "number" : 19 }, { "_id" : "tennis",
268
"number" : 18 }
269
270
CHAPTER 28
Aggregation Framework Reference
New in version 2.1.0. The aggregation framework provides the ability to project, process, and/or control the output of the query, without using map-reduce. Aggregation uses a syntax that resembles the same syntax and form as regular MongoDB database queries. These aggregation operations are all accessible by way of the aggregate() (page 918) method. While all examples in this document use this method, aggregate() (page 918) is merely a wrapper around the database command aggregate (page 808). The following prototype aggregation operations are equivalent:
db.people.aggregate( <pipeline> ) db.people.aggregate( [<pipeline>] ) db.runCommand( { aggregate: "people", pipeline: [<pipeline>] } )
These operations perform aggregation routines on the collection named people. <pipeline> is a placeholder for the aggregation pipeline denition. aggregate() (page 918) accepts the stages of the pipeline (i.e. <pipeline>) as an array, or as arguments to the method. This documentation provides an overview of all aggregation operators available for use in the aggregation pipeline as well as details regarding their use and behavior. See also: Aggregation Framework (page 255) overview, the Aggregation Framework Documentation Index (page 253), and the Aggregation Framework Examples (page 261) for more information on the aggregation functionality. Aggregation Operators: Pipeline (page 291) Expressions (page 300) $group Operators (page 300) Boolean Operators (page 302) Comparison Operators (page 303) Arithmetic Operators (page 304) String Operators (page 304) Date Operators (page 305) Conditional Expressions (page 306)
271
28.1 $add (aggregation)

$add Takes an array of one or more numbers and adds them together, returning the sum.
28.2 $addToSet (aggregation)

$addToSet Returns an array of all the values found in the selected eld among the documents in that group. Every unique value only appears once in the result set. There is no ordering guarantee for the output documents.
28.3 $and (aggregation)

$and Takes an array one or more values and returns true if all of the values in the array are true. Otherwise $and (page 272) returns false. Note: $and (page 272) uses short-circuit logic: the operation stops evaluation after encountering the rst false expression.
28.4 $avg (aggregation)

$avg Returns the average of all the values of the eld in all documents selected by this group.
28.5 $cmp (aggregation)

$cmp Takes two values in an array and returns an integer. The returned value is: A negative number if the rst value is less than the second. A positive number if the rst value is greater than the second. 0 if the two values are equal.
28.6 $concat (aggregation)

$concat New in version 2.4. Takes an array of strings, concatenates the strings, and returns the concatenated string. $concat (page 272) can only accept an array of strings. Use $concat (page 272) with the following syntax:
272
Chapter 28. Aggregation Framework Reference
{ $concat: [ <string>, <string>, ... ] }
If array element has a value of null or refers to a eld that is missing, $concat (page 272) will return null. Example Project new concatenated values. A collection menu contains the documents that stores information on menu items separately in the section, the category and the type elds, as in the following:
{ { { { _id: _id: _id: _id: 1, 2, 3, 4, item: item: item: item: { { { { sec: sec: sec: sec: "dessert", category: "pie", type: "apple" } } "dessert", category: "pie", type: "cherry" } } "main", category: "pie", type: "shepherds" } } "main", category: "pie", type: "chicken pot" } }
The following operation uses $concat (page 272) to concatenate the type eld from the sub-document item, a space, and the category eld from the sub-document item to project a new food eld:
db.menu.aggregate( { $project: { food: { $concat: [ "$item.type", " ", "$item.category" ] } } } )
The operation returns the following result set where the food eld contains the concatenated strings:
{ "result" : [ { { { { ], "ok" : 1 } "_id" "_id" "_id" "_id" : : : : 1, 2, 3, 4, "food" "food" "food" "food" : : : : "apple pie" }, "cherry pie" }, "shepherds pie" }, "chicken pot pie" }
Example Group by a concatenated string. A collection menu contains the documents that stores information on menu items separately in the section, the category and the type elds, as in the following:
{ { { { _id: _id: _id: _id: 1, 2, 3, 4, item: item: item: item: { { { { sec: sec: sec: sec: "dessert", category: "pie", type: "apple" } } "dessert", category: "pie", type: "cherry" } } "main", category: "pie", type: "shepherds" } } "main", category: "pie", type: "chicken pot" } }
The following aggregation uses $concat (page 272) to concatenate the sec eld from the sub-document item, the string ": ", and the category eld from the sub-document item to group by the new concatenated string and perform a count:
db.menu.aggregate( { $group: { _id: { $concat: [ "$item.sec",
28.6. $concat (aggregation)
273
": ", "$item.category" ] }, count: { $sum: 1 } } } )
The aggregation returns the following document:

{ "result" : [ { "_id" : "main: pie", "count" : 2 }, { "_id" : "dessert: pie", "count" : 2 } ], "ok" : 1 }
Example Concatenate null or missing values. A collection menu contains the documents that stores information on menu items separately in the section, the category and the type elds. Not all documents have the all three elds. For example, the document with _id equal to 5 is missing the category eld:
{ { { { { _id: _id: _id: _id: _id: 1, 2, 3, 4, 5, item: item: item: item: item: { { { { { sec: sec: sec: sec: sec: "dessert", category: "pie", type: "apple" } } "dessert", category: "pie", type: "cherry" } } "main", category: "pie", type: "shepherds" } } "main", category: "pie", type: "chicken pot" } } "beverage", type: "coffee" } }
The following aggregation uses the $concat (page 272) to concatenate the type eld from the sub-document item, a space, and the category eld from the sub-document item:
db.menu.aggregate( { $project: { food: { $concat: [ "$item.type", " ", "$item.category" ] } } } )
Because the document with _id equal to 5 is missing the type eld in the item sub-document, $concat (page 272) returns the value null as the concatenated value for the document:
{ "result" : [ { { { { { ], "ok" : 1 } "_id" "_id" "_id" "_id" "_id" : : : : : 1, 2, 3, 4, 5, "food" "food" "food" "food" "food" : : : : : "apple pie" }, "cherry pie" }, "shepherds pie" }, "chicken pot pie" }, null }
274
To handle possible missing elds, you can use $ifNull (page 280) with $concat (page 272), as in the following example which substitutes <unknown type> if the eld type is null or missing, and <unknown category> if the eld category is null or is missing:
db.menu.aggregate( { $project: { food:
{ $concat: [ { $ifNull: ["$item.type", "<unknown type>"] " ", { $ifNull: ["$item.category", "<unknown cate ] } } } )
The aggregation returns the following result set:

{ "result" : [ { { { { { ], "ok" : 1 } "_id" "_id" "_id" "_id" "_id" : : : : : 1, 2, 3, 4, 5, "food" "food" "food" "food" "food" : : : : : "apple pie" }, "cherry pie" }, "shepherds pie" }, "chicken pot pie" }, "coffee <unknown category>" }
28.7 $cond (aggregation)

$cond Use the $cond (page 275) operator with the following syntax:
{ $cond: [ <boolean-expression>, <true-case>, <false-case> ] }
Takes an array with three expressions, where the rst expression evaluates to a Boolean value. If the rst expression evaluates to true, $cond (page 275) returns the value of the second expression. If the rst expression evaluates to false, $cond (page 275) evaluates and returns the third expression.
28.8 $dayOfMonth (aggregation)

$dayOfMonth Takes a date and returns the day of the month as a number between 1 and 31.
28.9 $dayOfWeek (aggregation)

$dayOfWeek Takes a date and returns the day of the week as a number between 1 (Sunday) and 7 (Saturday.)
28.7. $cond (aggregation)
275
28.10 $dayOfYear (aggregation)

$dayOfYear Takes a date and returns the day of the year as a number between 1 and 366.
28.11 $divide (aggregation)

$divide Takes an array that contains a pair of numbers and returns the value of the rst number divided by the second number.
28.12 $eq (aggregation)

$eq Takes two values in an array and returns a boolean. The returned value is: true when the values are equivalent. false when the values are not equivalent.
28.13 $rst (aggregation)

$first Returns the rst value it encounters for its group . Note: Only use $first (page 276) when the $group (page 278) follows an $sort (page 287) operation. Otherwise, the result of this operation is unpredictable.
28.14 $geoNear (aggregation)

$geoNear New in version 2.4. $geoNear (page 276) returns documents in order of nearest to farthest from a specied point and pass the documents through the aggregation pipeline. Important: You can only use $geoNear (page 276) as the rst stage of a pipeline. You must include the distanceField option. The distanceField option species the eld that will contain the calculated distance. The collection must have a geospatial index (page 369). The $geoNear (page 276) accept the following options: Fields
276
near (coordinates) Species the coordinates (e.g. [ x, y ]) to use as the center of a geospatial query. distanceField (string) Species the output eld that will contain the calculated distance. You can use the dot notation to specify a eld within a subdocument. limit (number) Optional. Species the maximum number of documents to return. The default value is 100. See also the num option. num (number) Optional. Synonym for the limit option. If both num and limit are included, the num value overrides the limit value. maxDistance (number) Optional. Limits the results to the documents within the specied distance from the center coordinates. query (document) Optional. Limits the results to the documents that match the query. The query syntax is identical to the read operation query (page 168) syntax. spherical (boolean) Optional. Default value is false. When true, MongoDB performs calculation using spherical geometry. distanceMultiplier (number) Optional. Species a factor to multiply all distances returned by $geoNear (page 276). For example, use distanceMultiplier to convert from spherical queries returned in radians to linear units (i.e. miles or kilometers) by multiplying by the radius of the Earth. includeLocs (string) Optional. Species the output eld that identies the location used to calculate the distance. This option is useful when a location eld contains multiple locations. You can use the dot notation to specify a eld within a subdocument. uniqueDocs (boolean) Optional. Default value is false. If a location eld contains multiple locations, the default settings will return the document multiple times if more than one location meets the criteria. When true, the document will only return once even if the document has multiple locations that meet the criteria. Example The following aggregation nds at most 5 unique documents with a location at most .008 from the center [40.72, -73.99] and have type equal to public:
db.places.aggregate([ { $geoNear: { near: [40.724, -73.997], distanceField: "dist.calculated", maxDistance: 0.008, query: { type: "public" }, includeLocs: "dist.location", uniqueDocs: true, num: 5 } } ])
The aggregation returns the following:

{ "result" : [ { "_id" : 7, "name" : "Washington Square",
28.14. $geoNear (aggregation)
277
"type" : "public", "location" : [ [ 40.731, -73.999 ], [ 40.732, -73.998 ], [ 40.730, -73.995 ], [ 40.729, -73.996 ] ], "dist" : { "calculated" : 0.0050990195135962296, "location" : [ 40.729, -73.996 ] } }, { "_id" : 8, "name" : "Sara D. Roosevelt Park", "type" : "public", "location" : [ [ 40.723, -73.991 ], [ 40.723, -73.990 ], [ 40.715, -73.994 ], [ 40.715, -73.994 ] ], "dist" : { "calculated" : 0.006082762530298062, "location" : [ 40.723, -73.991 ] } } ], "ok" : 1 }
The matching documents in the result eld contain two new elds: dist.calculated eld that contains the calculated distance, and dist.location eld that contains the location used in the calculation. Note: The options for $geoNear (page 276) are similar to the geoNear (page 822) command with the following exceptions: distanceField is a mandatory eld for the $geoNear (page 276) pipeline operator; the option does not exist in the geoNear (page 822) command. includeLocs accepts a string in the $geoNear (page 276) pipeline operator and a boolean in the geoNear (page 822) command.
28.15 $group (aggregation)

$group Groups documents together for the purpose of calculating aggregate values based on a collection of documents. Practically, group often supports tasks such as average page views for each page in a website on a daily basis. The output of $group (page 278) depends on how you dene groups. Begin by specifying an identier (i.e. a _id eld) for the group youre creating with this pipeline. You can specify a single eld from the documents in the pipeline, a previously computed value, or an aggregate key made up from several incoming elds. Aggregate keys may resemble the following document:
278
{ _id : { author: $author, pageViews: $pageViews, posted: $posted } }
With the exception of the _id eld, $group (page 278) cannot output nested documents. Every group expression must specify an _id eld. You may specify the _id eld as a dotted eld path reference, a document with multiple elds enclosed in braces (i.e. { and }), or a constant value. Note: Use $project (page 285) as needed to rename the grouped eld after an $group (page 278) operation, if necessary. Consider the following example:
db.article.aggregate( { $group : { _id : "$author", docsPerAuthor : { $sum : 1 }, viewsPerAuthor : { $sum : "$pageViews" } }} );
This groups by the author eld and computes two elds, the rst docsPerAuthor is a counter eld that adds one for each document with a given author eld using the $sum (page 289) function. The viewsPerAuthor eld is the sum of all of the pageViews elds in the documents for each group. Each eld dened for the $group (page 278) must use one of the group aggregation function listed below to generate its composite value: $addToSet (page 272) $first (page 276) $last (page 280) $max (page 282) $min (page 283) $avg (page 272) $push (page 287) $sum (page 289) Warning: The aggregation system currently stores $group (page 278) operations in memory, which may cause problems when processing a larger number of groups.
28.16 $gt (aggregation)

$gt Takes two values in an array and returns a boolean. The returned value is: true when the rst value is greater than the second value. false when the rst value is less than or equal to the second value.
28.16. $gt (aggregation)
279
28.17 $gte (aggregation)

$gte Takes two values in an array and returns a boolean. The returned value is: true when the rst value is greater than or equal to the second value. false when the rst value is less than the second value.
28.18 $hour (aggregation)

$hour Takes a date and returns the hour between 0 and 23.
28.19 $ifNull (aggregation)

$ifNull Use the $ifNull (page 280) operator with the following syntax:
{ $ifNull: [ <expression>, <replacement-if-null> ] }
Takes an array with two expressions. $ifNull (page 280) returns the rst expression if it evaluates to a non-null value. Otherwise, $ifNull (page 280) returns the second expressions value.
28.20 $last (aggregation)

$last Returns the last value it encounters for its group. Note: Only use $last (page 280) when the $group (page 278) follows an $sort (page 287) operation. Otherwise, the result of this operation is unpredictable.
28.21 $limit (aggregation)

$limit Restricts the number of documents that pass through the $limit (page 280) in the pipeline. $limit (page 280) takes a single numeric (positive whole number) value as a parameter. Once the specied number of documents pass through the pipeline operator, no more will. Consider the following example:
db.article.aggregate( { $limit : 5 } );
This operation returns only the rst 5 documents passed to it from by the pipeline. $limit (page 280) has no effect on the content of the documents it passes. Note: Changed in version 2.4: When a $sort (page 287) immediately precedes a $limit (page 280) in the pipeline, the $sort (page 287) operation only maintains the top n results as it progresses, where n is the
280
specied limit. Before 2.4, $sort (page 287) would sort all the results in memory, and then limit the results to n results.
28.22 $lt (aggregation)

$lt Takes two values in an array and returns a boolean. The returned value is: true when the rst value is less than the second value. false when the rst value is greater than or equal to the second value.
28.23 $lte (aggregation)

$lte Takes two values in an array and returns a boolean. The returned value is: true when the rst value is less than or equal to the second value. false when the rst value is greater than the second value.
28.24 $match (aggregation)

$match $match (page 281) pipes the documents that match its conditions to the next operator in the pipeline. The $match (page 281) query syntax is identical to the read operation query (page 168) syntax. Example The following operation uses $match (page 281) to perform a simple equality match:
db.articles.aggregate( { $match : { author : "dave" } } );
The $match (page 281) selects the documents where the author eld equals dave, and the aggregation returns the following:
{ "result" : [ { "_id" : ObjectId("512bc95fe835e68f199c8686"), "author": "dave", "score" : 80 }, { "_id" : ObjectId("512bc962e835e68f199c8687"), "author" : "dave", "score" : 85 } ], "ok" : 1 }
28.22. $lt (aggregation)
281
Example The following example selects documents to process using the $match (page 281) pipeline operator and then pipes the results to the $group (page 278) pipeline operator to compute a count of the documents:
db.articles.aggregate( [ { $match : { score : { $gt : 70, $lte : 90 } } }, { $group: { _id: null, count: { $sum: 1 } } } ] );
In the aggregation pipeline, $match (page 281) selects the documents where the score is greater than 70 and less than or equal to 90. These documents are then piped to the $group (page 278) to perform a count. The aggregation returns the following:
{ "result" : [ { "_id" : null, "count" : 3 } ], "ok" : 1 }
Note: Place the $match (page 281) as early in the aggregation pipeline as possible. Because $match (page 281) limits the total number of documents in the aggregation pipeline, earlier $match (page 281) operations minimize the amount of processing down the pipe. If you place a $match (page 281) at the very beginning of a pipeline, the query can take advantage of indexes like any other db.collection.find() (page 924) or db.collection.findOne() (page 929). New in version 2.4: $match (page 281) queries can support the geospatial $geoWithin (page 778) operations. Warning: You cannot use $where (page 775) in $match (page 281) queries as part of the aggregation pipeline.
28.25 $max (aggregation)

$max Returns the highest value among all values of the eld in all documents selected by this group.
28.26 $millisecond (aggregation)

$millisecond Takes a date and returns the millisecond portion of the date as an integer between 0 and 999.
282
28.27 $min (aggregation)

$min The $min (page 283) operator returns the lowest non-null value of a eld in the documents for a $group (page 278) operation. Changed in version 2.4: If some, but not all, documents for the $min (page 283) operation have either a null value for the eld or are missing the eld, the $min (page 283) operator only considers the non-null and the non-missing values for the eld. If all documents for the $min (page 283) operation have null value for the eld or are missing the eld, the $min (page 283) operator returns null for the minimum value. Before 2.4, if any of the documents for the $min (page 283) operation were missing the eld, the $min (page 283) operator would not return any value. If any of the documents for the $min (page 283) had the value null, the $min (page 283) operator would return a null. Example The users collection contains the following documents:
{ { { { "_id" "_id" "_id" "_id" : : : : "abc001", "abe001", "efg001", "xyz001", "age" "age" "age" "age" : : : : 25 35 20 15 } } } }
To nd the minimum value of the age eld from all the documents, use the $min (page 283) operator:
db.users.aggregate( [ { $group: { _id:0, minAge: { $min: "$age"} } } ] )
The operation returns the value of the age eld in the minAge eld:
{ "result" : [ { "_id" : 0, "minAge" : 15 } ], "ok" : 1 }
To nd the minimum value of the age eld for only those documents with _id starting with the letter a, use the $min (page 283) operator after a $match (page 281) operation:
db.users.aggregate( [ { $match: { _id: /â/ } }, { $group: { _id: 0, minAge: { $min: "$age"} } } ] )
The operation returns the minimum value of the age eld for the two documents with _id starting with the letter a:
Example The users collection contains the following documents where some of the documents are either missing the age eld or the age eld contains null:
{ { { { { { "_id" "_id" "_id" "_id" "_id" "_id" : : : : : : "abc001", "age" "abe001", "age" "efg001", "age" "xyz001", "age" "xxx001" } "zzz001", "age" : : : : 25 35 20 15 } } } }
: null }
The following operation nds the minimum value of the age eld in all the documents:
28.27. $min (aggregation)
283
Because only some documents for the $min (page 283) operation are missing the age eld or have age eld equal to null, $min (page 283) only considers the non-null and the non-missing values and the operation returns the following document:
The following operation nds the minimum value of the age eld for only those documents where the _id equals "xxx001" or "zzz001":
db.users.aggregate( [ { $match: { _id: {$in: [ "xxx001", "zzz001" ] } } }, { $group: { _id: 0, minAge: { $min: "$age"} } } ] )
The $min (page 283) operation returns null for the minimum age since all documents for the $min (page 283) operation have null value for the eld age or are missing the eld:
{ "result" : [ { "_id" : 0, "minAge" : null } ], "ok" : 1 }
28.28 $minute (aggregation)

$minute Takes a date and returns the minute between 0 and 59.
28.29 $mod (aggregation)

$mod Takes an array that contains a pair of numbers and returns the remainder of the rst number divided by the second number. See also: $mod (page 772)
28.30 $month (aggregation)

$month Takes a date and returns the month as a number between 1 and 12.
28.31 $multiply (aggregation)

$multiply Takes an array of one or more numbers and multiples them, returning the resulting product.
284
28.32 $ne (aggregation)

$ne Takes two values in an array returns a boolean. The returned value is: true when the values are not equivalent. false when the values are equivalent.
28.33 $not (aggregation)

$not Returns the boolean opposite value passed to it. When passed a true value, $not (page 285) returns false; when passed a false value, $not (page 285) returns true.
28.34 $or (aggregation)

$or Takes an array of one or more values and returns true if any of the values in the array are true. Otherwise $or (page 285) returns false. Note: $or (page 285) uses short-circuit logic: the operation stops evaluation after encountering the rst true expression.
28.35 $project (aggregation)

$project Reshapes a document stream by renaming, adding, or removing elds. Also use $project (page 285) to create computed values or sub-objects. Use $project (page 285) to: Include elds from the original document. Insert computed elds. Rename elds. Create and populate elds that hold sub-documents. Use $project (page 285) to quickly select the elds that you want to include or exclude from the response. Consider the following aggregation framework operation.
db.article.aggregate( { $project : { title : 1 , author : 1 , }} );
This operation includes the title eld and the author eld in the document that returns from the aggregation pipeline. Note: The _id eld is always included by default. You may explicitly exclude _id as follows:
28.32. $ne (aggregation)
285
db.article.aggregate( { $project : { _id : 0 , title : 1 , author : 1 }} );
Here, the projection excludes the _id eld but includes the title and author elds. Projections can also add computed elds to the document stream passing through the pipeline. A computed eld can use any of the expression operators (page 300). Consider the following example:
db.article.aggregate( { $project : { title : 1, doctoredPageViews : { $add:["$pageViews", 10] } }} );
Here, the eld doctoredPageViews represents the value of the pageViews eld after adding 10 to the original eld using the $add (page 272). Note: You must enclose the expression that denes the computed eld in braces, so that the expression is a valid object. You may also use $project (page 285) to rename elds. Consider the following example:
db.article.aggregate( { $project : { title : 1 , page_views : "$pageViews" , bar : "$other.foo" }} );
This operation renames the pageViews eld to page_views, and renames the foo eld in the other subdocument as the top-level eld bar. The eld references used for renaming elds are direct expressions and do not use an operator or surrounding braces. All aggregation eld references can use dotted paths to refer to elds in nested documents. Finally, you can use the $project (page 285) to create and populate new sub-documents. Consider the following example that creates a new object-valued eld named stats that holds a number of values:
db.article.aggregate( { $project : { title : 1 , stats : { pv : "$pageViews", foo : "$other.foo", dpv : { $add:["$pageViews", 10] } } }} );
This projection includes the title eld and places $project (page 285) into inclusive mode. Then, it creates the stats documents with the following elds: pv which includes and renames the pageViews from the top level of the original documents. 286 Chapter 28. Aggregation Framework Reference
foo which includes the value of other.foo from the original documents. dpv which is a computed eld that adds 10 to the value of the pageViews eld in the original document using the $add (page 272) aggregation expression.
28.36 $push (aggregation)

$push Returns an array of all the values found in the selected eld among the documents in that group. A value may appear more than once in the result set if more than one eld in the grouped documents has that value.
28.37 $second (aggregation)

$second Takes a date and returns the second between 0 and 59, but can be 60 to account for leap seconds.
28.38 $skip (aggregation)

$skip Skips over the specied number of documents that pass through the $skip (page 287) in the pipeline before passing all of the remaining input. $skip (page 287) takes a single numeric (positive whole number) value as a parameter. Once the operation has skipped the specied number of documents, it passes all the remaining documents along the pipeline without alteration. Consider the following example:
db.article.aggregate( { $skip : 5 } );
This operation skips the rst 5 documents passed to it by the pipeline. $skip (page 287) has no effect on the content of the documents it passes along the pipeline.
28.39 $sort (aggregation)

$sort The $sort (page 287) pipeline operator sorts all input documents and returns them to the pipeline in sorted order. Consider the following prototype form:
db.<collection-name>.aggregate( { $sort : { <sort-key> } } );
This sorts the documents in the collection named <collection-name>, according to the key and specication in the { <sort-key> } document. Specify the sort in a document with a eld or elds that you want to sort by and a value of 1 or -1 to specify an ascending or descending sort respectively, as in the following example:
28.36. $push (aggregation)
287
db.users.aggregate( { $sort : { age : -1, posts: 1 } } );
This operation sorts the documents in the users collection, in descending order according by the age eld and then in ascending order according to the value in the posts eld. When comparing values of different BSON types, MongoDB uses the following comparison order, from lowest to highest: 1.MinKey (internal type) 2.Null 3.Numbers (ints, longs, doubles) 4.Symbol, String 5.Object 6.Array 7.BinData 8.ObjectID 9.Boolean 10.Date, Timestamp 11.Regular Expression 12.MaxKey (internal type) Note: MongoDB treats some types as equivalent for comparison purposes. For instance, numeric types undergo conversion before comparison. Note: The $sort (page 287) cannot begin sorting documents until previous operators in the pipeline have returned all output. $skip (page 287) $sort (page 287) operator can take advantage of an index when placed at the beginning of the pipeline or placed before the following aggregation operators: $project (page 285) $unwind (page 290) $group (page 278). Changed in version 2.4: When a $sort (page 287) immediately precedes a $limit (page 280) in the pipeline, the $sort (page 287) operation only maintains the top n results as it progresses, where n is the specied limit. Before 2.4, $sort (page 287) would sort all the results in memory, and then limit the results to n results. Warning: Changed in version 2.4: Sorts immediately proceeded by a limit no longer need to t into memory. Previously, all sorts had to t into memory or use an index. Unless the $sort (page 287) operator can use an index, or immediately precedes a $limit (page 280), the $sort (page 287) operation must t within memory. For $sort (page 287) operations that immediately precede a $limit (page 280) stage, MongoDB only needs to store the number of items specied by $limit (page 280) in memory.
288
28.40 $strcasecmp (aggregation)

$strcasecmp Takes in two strings. Returns a number. $strcasecmp (page 289) is positive if the rst string is greater than the second and negative if the rst string is less than the second. $strcasecmp (page 289) returns 0 if the strings are identical. Note: $strcasecmp (page 289) may not make sense when applied to glyphs outside the Roman alphabet. $strcasecmp (page 289) internally capitalizes strings before comparing them to provide a case-insensitive comparison. Use $cmp (page 272) for a case sensitive comparison.
28.41 $substr (aggregation)

$substr $substr (page 289) takes a string and two numbers. The rst number represents the number of bytes in the string to skip, and the second number species the number of bytes to return from the string. Note: $substr (page 289) is not encoding aware and if used improperly may produce a result string containing an invalid UTF-8 character sequence.
28.42 $subtract (aggregation)

$subtract Takes an array that contains a pair of numbers and subtracts the second from the rst, returning their difference.
28.43 $sum (aggregation)

$sum Returns the sum of all the values for a specied eld in the grouped documents, as in the second use above. Alternately, if you specify a value as an argument, $sum (page 289) will increment this eld by the specied value for every document in the grouping. Typically, as in the rst use above, specify a value of 1 in order to count members of the group.
28.44 $toLower (aggregation)

$toLower Takes a single string and converts that string to lowercase, returning the result. All uppercase letters become lowercase. Note: $toLower (page 289) may not make sense when applied to glyphs outside the Roman alphabet.
28.40. $strcasecmp (aggregation)
289
28.45 $toUpper (aggregation)

$toUpper Takes a single string and converts that string to uppercase, returning the result. All lowercase letters become uppercase. Note: $toUpper (page 290) may not make sense when applied to glyphs outside the Roman alphabet.
28.46 $unwind (aggregation)

$unwind Peels off the elements of an array individually, and returns a stream of documents. $unwind (page 290) returns one document for every member of the unwound array within every source document. Take the following aggregation command:
db.article.aggregate( { $project : { author : 1 , title : 1 , tags : 1 }}, { $unwind : "$tags" } );
Note: The dollar sign (i.e. $) must proceed the eld specication handed to the $unwind (page 290) operator. In the above aggregation $project (page 285) selects (inclusively) the author, title, and tags elds, as well as the _id eld implicitly. Then the pipeline passes the results of the projection to the $unwind (page 290) operator, which will unwind the tags eld. This operation may return a sequence of documents that resemble the following for a collection that contains one document holding a tags eld with an array of 3 items.
{ "result" : [ { "_id" : ObjectId("4e6e4ef557b77501a49233f6"), "title" : "this is my title", "author" : "bob", "tags" : "fun" }, { "_id" : ObjectId("4e6e4ef557b77501a49233f6"), "title" : "this is my title", "author" : "bob", "tags" : "good" }, { "_id" : ObjectId("4e6e4ef557b77501a49233f6"), "title" : "this is my title", "author" : "bob", "tags" : "fun" } ],
290
"OK" : 1 }
A single document becomes 3 documents: each document is identical except for the value of the tags eld. Each value of tags is one of the values in the original tags array. Note: $unwind (page 290) has the following behaviors: $unwind (page 290) is most useful in combination with $group (page 278). You may undo the effects of unwind operation with the $group (page 278) pipeline operator. If you specify a target eld for $unwind (page 290) that does not exist in an input document, the pipeline ignores the input document, and will generate no result documents. If you specify a target eld for $unwind (page 290) db.collection.aggregate() (page 918) generates an error. that is not an array,
If you specify a target eld for $unwind (page 290) that holds an empty array ([]) in an input document, the pipeline ignores the input document, and will generates no result documents.
28.47 $week (aggregation)

$week Takes a date and returns the week of the year as a number between 0 and 53. Weeks begin on Sundays, and week 1 begins with the rst Sunday of the year. Days preceding the rst Sunday of the year are in week 0. This behavior is the same as the %U operator to the strftime standard library function.
28.48 $year (aggregation)

$year Takes a date and returns the full year.
28.49 Pipeline
Warning: The pipeline cannot operate on values of the following types: Binary, Symbol, MinKey, MaxKey, DBRef, Code, and CodeWScope. Pipeline operators appear in an array. Conceptually, documents pass through these operators in a sequence. All examples in this section assume that the aggregation pipeline begins with a collection named article that contains documents that resemble the following:
{ title : "this is my title" , author : "bob" , posted : new Date() , pageViews : 5 , tags : [ "fun" , "good" , "fun" ] , comments : [
28.47. $week (aggregation)
291
{ author :"joe" , text : "this is cool" } , { author :"sam" , text : "this is bad" } ], other : { foo : 5 } }
The current pipeline operators are: $project Reshapes a document stream by renaming, adding, or removing elds. Also use $project (page 285) to create computed values or sub-objects. Use $project (page 285) to: Include elds from the original document. Insert computed elds. Rename elds. Create and populate elds that hold sub-documents. Use $project (page 285) to quickly select the elds that you want to include or exclude from the response. Consider the following aggregation framework operation.
db.article.aggregate( { $project : { title : 1 , author : 1 , }} );
This operation includes the title eld and the author eld in the document that returns from the aggregation pipeline. Note: The _id eld is always included by default. You may explicitly exclude _id as follows:
db.article.aggregate( { $project : { _id : 0 , title : 1 , author : 1 }} );
Here, the projection excludes the _id eld but includes the title and author elds. Projections can also add computed elds to the document stream passing through the pipeline. A computed eld can use any of the expression operators (page 300). Consider the following example:
db.article.aggregate( { $project : { title : 1, doctoredPageViews : { $add:["$pageViews", 10] } }} );
Here, the eld doctoredPageViews represents the value of the pageViews eld after adding 10 to the original eld using the $add (page 272). Note: You must enclose the expression that denes the computed eld in braces, so that the expression is a valid object.
292
You may also use $project (page 285) to rename elds. Consider the following example:
db.article.aggregate( { $project : { title : 1 , page_views : "$pageViews" , bar : "$other.foo" }} );
This operation renames the pageViews eld to page_views, and renames the foo eld in the other subdocument as the top-level eld bar. The eld references used for renaming elds are direct expressions and do not use an operator or surrounding braces. All aggregation eld references can use dotted paths to refer to elds in nested documents. Finally, you can use the $project (page 285) to create and populate new sub-documents. Consider the following example that creates a new object-valued eld named stats that holds a number of values:
db.article.aggregate( { $project : { title : 1 , stats : { pv : "$pageViews", foo : "$other.foo", dpv : { $add:["$pageViews", 10] } } }} );
This projection includes the title eld and places $project (page 285) into inclusive mode. Then, it creates the stats documents with the following elds: pv which includes and renames the pageViews from the top level of the original documents. foo which includes the value of other.foo from the original documents. dpv which is a computed eld that adds 10 to the value of the pageViews eld in the original document using the $add (page 272) aggregation expression. $match $match (page 281) pipes the documents that match its conditions to the next operator in the pipeline. The $match (page 281) query syntax is identical to the read operation query (page 168) syntax. Example The following operation uses $match (page 281) to perform a simple equality match:
db.articles.aggregate( { $match : { author : "dave" } } );
The $match (page 281) selects the documents where the author eld equals dave, and the aggregation returns the following:
{ "result" : [ { "_id" : ObjectId("512bc95fe835e68f199c8686"), "author": "dave", "score" : 80
28.49. Pipeline
293
}, { "_id" : ObjectId("512bc962e835e68f199c8687"), "author" : "dave", "score" : 85 } ], "ok" : 1 }
Example The following example selects documents to process using the $match (page 281) pipeline operator and then pipes the results to the $group (page 278) pipeline operator to compute a count of the documents:
db.articles.aggregate( [ { $match : { score : { $gt : 70, $lte : 90 } } }, { $group: { _id: null, count: { $sum: 1 } } } ] );
In the aggregation pipeline, $match (page 281) selects the documents where the score is greater than 70 and less than or equal to 90. These documents are then piped to the $group (page 278) to perform a count. The aggregation returns the following:
{ "result" : [ { "_id" : null, "count" : 3 } ], "ok" : 1 }
Note: Place the $match (page 281) as early in the aggregation pipeline as possible. Because $match (page 281) limits the total number of documents in the aggregation pipeline, earlier $match (page 281) operations minimize the amount of processing down the pipe. If you place a $match (page 281) at the very beginning of a pipeline, the query can take advantage of indexes like any other db.collection.find() (page 924) or db.collection.findOne() (page 929). New in version 2.4: $match (page 281) queries can support the geospatial $geoWithin (page 778) operations. Warning: You cannot use $where (page 775) in $match (page 281) queries as part of the aggregation pipeline. $limit Restricts the number of documents that pass through the $limit (page 280) in the pipeline. $limit (page 280) takes a single numeric (positive whole number) value as a parameter. Once the specied number of documents pass through the pipeline operator, no more will. Consider the following example:
db.article.aggregate( { $limit : 5 } );
294
This operation returns only the rst 5 documents passed to it from by the pipeline. $limit (page 280) has no effect on the content of the documents it passes. Note: Changed in version 2.4: When a $sort (page 287) immediately precedes a $limit (page 280) in the pipeline, the $sort (page 287) operation only maintains the top n results as it progresses, where n is the specied limit. Before 2.4, $sort (page 287) would sort all the results in memory, and then limit the results to n results. $skip Skips over the specied number of documents that pass through the $skip (page 287) in the pipeline before passing all of the remaining input. $skip (page 287) takes a single numeric (positive whole number) value as a parameter. Once the operation has skipped the specied number of documents, it passes all the remaining documents along the pipeline without alteration. Consider the following example:
db.article.aggregate( { $skip : 5 } );
This operation skips the rst 5 documents passed to it by the pipeline. $skip (page 287) has no effect on the content of the documents it passes along the pipeline. $unwind Peels off the elements of an array individually, and returns a stream of documents. $unwind (page 290) returns one document for every member of the unwound array within every source document. Take the following aggregation command:
db.article.aggregate( { $project : { author : 1 , title : 1 , tags : 1 }}, { $unwind : "$tags" } );
Note: The dollar sign (i.e. $) must proceed the eld specication handed to the $unwind (page 290) operator. In the above aggregation $project (page 285) selects (inclusively) the author, title, and tags elds, as well as the _id eld implicitly. Then the pipeline passes the results of the projection to the $unwind (page 290) operator, which will unwind the tags eld. This operation may return a sequence of documents that resemble the following for a collection that contains one document holding a tags eld with an array of 3 items.
{ "result" : [ { "_id" : ObjectId("4e6e4ef557b77501a49233f6"), "title" : "this is my title", "author" : "bob", "tags" : "fun" }, { "_id" : ObjectId("4e6e4ef557b77501a49233f6"), "title" : "this is my title", "author" : "bob",
28.49. Pipeline
295
"tags" : "good" }, { "_id" : ObjectId("4e6e4ef557b77501a49233f6"), "title" : "this is my title", "author" : "bob", "tags" : "fun" } ], "OK" : 1 }
A single document becomes 3 documents: each document is identical except for the value of the tags eld. Each value of tags is one of the values in the original tags array. Note: $unwind (page 290) has the following behaviors: $unwind (page 290) is most useful in combination with $group (page 278). You may undo the effects of unwind operation with the $group (page 278) pipeline operator. If you specify a target eld for $unwind (page 290) that does not exist in an input document, the pipeline ignores the input document, and will generate no result documents. If you specify a target eld for $unwind (page 290) db.collection.aggregate() (page 918) generates an error. that is not an array,
If you specify a target eld for $unwind (page 290) that holds an empty array ([]) in an input document, the pipeline ignores the input document, and will generates no result documents. $group Groups documents together for the purpose of calculating aggregate values based on a collection of documents. Practically, group often supports tasks such as average page views for each page in a website on a daily basis. The output of $group (page 278) depends on how you dene groups. Begin by specifying an identier (i.e. a _id eld) for the group youre creating with this pipeline. You can specify a single eld from the documents in the pipeline, a previously computed value, or an aggregate key made up from several incoming elds. Aggregate keys may resemble the following document:
{ _id : { author: $author, pageViews: $pageViews, posted: $posted } }
With the exception of the _id eld, $group (page 278) cannot output nested documents. Every group expression must specify an _id eld. You may specify the _id eld as a dotted eld path reference, a document with multiple elds enclosed in braces (i.e. { and }), or a constant value. Note: Use $project (page 285) as needed to rename the grouped eld after an $group (page 278) operation, if necessary. Consider the following example:
db.article.aggregate( { $group : { _id : "$author", docsPerAuthor : { $sum : 1 }, viewsPerAuthor : { $sum : "$pageViews" } }} );
296
This groups by the author eld and computes two elds, the rst docsPerAuthor is a counter eld that adds one for each document with a given author eld using the $sum (page 289) function. The viewsPerAuthor eld is the sum of all of the pageViews elds in the documents for each group. Each eld dened for the $group (page 278) must use one of the group aggregation function listed below to generate its composite value: $addToSet (page 272) $first (page 276) $last (page 280) $max (page 282) $min (page 283) $avg (page 272) $push (page 287) $sum (page 289) Warning: The aggregation system currently stores $group (page 278) operations in memory, which may cause problems when processing a larger number of groups. $sort The $sort (page 287) pipeline operator sorts all input documents and returns them to the pipeline in sorted order. Consider the following prototype form:
db.<collection-name>.aggregate( { $sort : { <sort-key> } } );
This sorts the documents in the collection named <collection-name>, according to the key and specication in the { <sort-key> } document. Specify the sort in a document with a eld or elds that you want to sort by and a value of 1 or -1 to specify an ascending or descending sort respectively, as in the following example:
db.users.aggregate( { $sort : { age : -1, posts: 1 } } );
This operation sorts the documents in the users collection, in descending order according by the age eld and then in ascending order according to the value in the posts eld. When comparing values of different BSON types, MongoDB uses the following comparison order, from lowest to highest: 1.MinKey (internal type) 2.Null 3.Numbers (ints, longs, doubles) 4.Symbol, String 5.Object 6.Array 7.BinData 8.ObjectID 28.49. Pipeline 297
9.Boolean 10.Date, Timestamp 11.Regular Expression 12.MaxKey (internal type) Note: MongoDB treats some types as equivalent for comparison purposes. For instance, numeric types undergo conversion before comparison. Note: The $sort (page 287) cannot begin sorting documents until previous operators in the pipeline have returned all output. $skip (page 287) $sort (page 287) operator can take advantage of an index when placed at the beginning of the pipeline or placed before the following aggregation operators: $project (page 285) $unwind (page 290) $group (page 278). Changed in version 2.4: When a $sort (page 287) immediately precedes a $limit (page 280) in the pipeline, the $sort (page 287) operation only maintains the top n results as it progresses, where n is the specied limit. Before 2.4, $sort (page 287) would sort all the results in memory, and then limit the results to n results. Warning: Changed in version 2.4: Sorts immediately proceeded by a limit no longer need to t into memory. Previously, all sorts had to t into memory or use an index. Unless the $sort (page 287) operator can use an index, or immediately precedes a $limit (page 280), the $sort (page 287) operation must t within memory. For $sort (page 287) operations that immediately precede a $limit (page 280) stage, MongoDB only needs to store the number of items specied by $limit (page 280) in memory. $geoNear New in version 2.4. $geoNear (page 276) returns documents in order of nearest to farthest from a specied point and pass the documents through the aggregation pipeline. Important: You can only use $geoNear (page 276) as the rst stage of a pipeline. You must include the distanceField option. The distanceField option species the eld that will contain the calculated distance. The collection must have a geospatial index (page 369). The $geoNear (page 276) accept the following options: Fields near (coordinates) Species the coordinates (e.g. [ x, y ]) to use as the center of a geospatial query. distanceField (string) Species the output eld that will contain the calculated distance. You can use the dot notation to specify a eld within a subdocument.
298
limit (number) Optional. Species the maximum number of documents to return. The default value is 100. See also the num option. num (number) Optional. Synonym for the limit option. If both num and limit are included, the num value overrides the limit value. maxDistance (number) Optional. Limits the results to the documents within the specied distance from the center coordinates. query (document) Optional. Limits the results to the documents that match the query. The query syntax is identical to the read operation query (page 168) syntax. spherical (boolean) Optional. Default value is false. When true, MongoDB performs calculation using spherical geometry. distanceMultiplier (number) Optional. Species a factor to multiply all distances returned by $geoNear (page 276). For example, use distanceMultiplier to convert from spherical queries returned in radians to linear units (i.e. miles or kilometers) by multiplying by the radius of the Earth. includeLocs (string) Optional. Species the output eld that identies the location used to calculate the distance. This option is useful when a location eld contains multiple locations. You can use the dot notation to specify a eld within a subdocument. uniqueDocs (boolean) Optional. Default value is false. If a location eld contains multiple locations, the default settings will return the document multiple times if more than one location meets the criteria. When true, the document will only return once even if the document has multiple locations that meet the criteria. Example The following aggregation nds at most 5 unique documents with a location at most .008 from the center [40.72, -73.99] and have type equal to public:
db.places.aggregate([ { $geoNear: { near: [40.724, -73.997], distanceField: "dist.calculated", maxDistance: 0.008, query: { type: "public" }, includeLocs: "dist.location", uniqueDocs: true, num: 5 } } ])
The aggregation returns the following:

{ "result" : [ { "_id" : 7, "name" : "Washington Square", "type" : "public", "location" : [ [ 40.731, -73.999 ], [ 40.732, -73.998 ], [ 40.730, -73.995 ],
28.49. Pipeline
299
[ 40.729, -73.996 ] ], "dist" : { "calculated" : 0.0050990195135962296, "location" : [ 40.729, -73.996 ] } }, { "_id" : 8, "name" : "Sara D. Roosevelt Park", "type" : "public", "location" : [ [ 40.723, -73.991 ], [ 40.723, -73.990 ], [ 40.715, -73.994 ], [ 40.715, -73.994 ] ], "dist" : { "calculated" : 0.006082762530298062, "location" : [ 40.723, -73.991 ] } } ], "ok" : 1 }
The matching documents in the result eld contain two new elds: dist.calculated eld that contains the calculated distance, and dist.location eld that contains the location used in the calculation. Note: The options for $geoNear (page 276) are similar to the geoNear (page 822) command with the following exceptions: distanceField is a mandatory eld for the $geoNear (page 276) pipeline operator; the option does not exist in the geoNear (page 822) command. includeLocs accepts a string in the $geoNear (page 276) pipeline operator and a boolean in the geoNear (page 822) command.
28.50 Expressions
These operators calculate values within the aggregation framework.
28.50.1 $group Operators

The $group (page 278) pipeline stage provides the following operations: $addToSet Returns an array of all the values found in the selected eld among the documents in that group. Every unique value only appears once in the result set. There is no ordering guarantee for the output documents. $first Returns the rst value it encounters for its group .
300
Note: Only use $first (page 276) when the $group (page 278) follows an $sort (page 287) operation. Otherwise, the result of this operation is unpredictable. $last Returns the last value it encounters for its group. Note: Only use $last (page 280) when the $group (page 278) follows an $sort (page 287) operation. Otherwise, the result of this operation is unpredictable. $max Returns the highest value among all values of the eld in all documents selected by this group. $min The $min (page 283) operator returns the lowest non-null value of a eld in the documents for a $group (page 278) operation. Changed in version 2.4: If some, but not all, documents for the $min (page 283) operation have either a null value for the eld or are missing the eld, the $min (page 283) operator only considers the non-null and the non-missing values for the eld. If all documents for the $min (page 283) operation have null value for the eld or are missing the eld, the $min (page 283) operator returns null for the minimum value. Before 2.4, if any of the documents for the $min (page 283) operation were missing the eld, the $min (page 283) operator would not return any value. If any of the documents for the $min (page 283) had the value null, the $min (page 283) operator would return a null. Example The users collection contains the following documents:
{ { { { "_id" "_id" "_id" "_id" : : : : "abc001", "abe001", "efg001", "xyz001", "age" "age" "age" "age" : : : : 25 35 20 15 } } } }
To nd the minimum value of the age eld from all the documents, use the $min (page 283) operator:
The operation returns the value of the age eld in the minAge eld:
To nd the minimum value of the age eld for only those documents with _id starting with the letter a, use the $min (page 283) operator after a $match (page 281) operation:
db.users.aggregate( [ { $match: { _id: /â/ } }, { $group: { _id: 0, minAge: { $min: "$age"} } } ] )
The operation returns the minimum value of the age eld for the two documents with _id starting with the letter a:
Example The users collection contains the following documents where some of the documents are either missing the age eld or the age eld contains null: 28.50. Expressions 301
{ { { { { {
"_id" "_id" "_id" "_id" "_id" "_id"
: : : : : :
"abc001", "age" "abe001", "age" "efg001", "age" "xyz001", "age" "xxx001" } "zzz001", "age"
: : : :
25 35 20 15
} } } }
: null }
The following operation nds the minimum value of the age eld in all the documents:
Because only some documents for the $min (page 283) operation are missing the age eld or have age eld equal to null, $min (page 283) only considers the non-null and the non-missing values and the operation returns the following document:
The following operation nds the minimum value of the age eld for only those documents where the _id equals "xxx001" or "zzz001":
db.users.aggregate( [ { $match: { _id: {$in: [ "xxx001", "zzz001" ] } } }, { $group: { _id: 0, minAge: { $min: "$age"} } } ] )
The $min (page 283) operation returns null for the minimum age since all documents for the $min (page 283) operation have null value for the eld age or are missing the eld:
{ "result" : [ { "_id" : 0, "minAge" : null } ], "ok" : 1 }
$avg Returns the average of all the values of the eld in all documents selected by this group. $push Returns an array of all the values found in the selected eld among the documents in that group. A value may appear more than once in the result set if more than one eld in the grouped documents has that value. $sum Returns the sum of all the values for a specied eld in the grouped documents, as in the second use above. Alternately, if you specify a value as an argument, $sum (page 289) will increment this eld by the specied value for every document in the grouping. Typically, as in the rst use above, specify a value of 1 in order to count members of the group.
28.50.2 Boolean Operators

The three boolean operators accept Booleans as arguments and return Booleans as results. Note: These operators convert non-booleans to Boolean values according to the BSON standards. Here, null, undefined, and 0 values become false, while non-zero numeric values, and all other types, such as strings, dates, objects become true. $and Takes an array one or more values and returns true if all of the values in the array are true. Otherwise $and (page 272) returns false.
302
Note: $and (page 272) uses short-circuit logic: the operation stops evaluation after encountering the rst false expression. $or Takes an array of one or more values and returns true if any of the values in the array are true. Otherwise $or (page 285) returns false. Note: $or (page 285) uses short-circuit logic: the operation stops evaluation after encountering the rst true expression. $not Returns the boolean opposite value passed to it. When passed a true value, $not (page 285) returns false; when passed a false value, $not (page 285) returns true.
28.50.3 Comparison Operators

These operators perform comparisons between two values and return a Boolean, in most cases, reecting the result of that comparison. All comparison operators take an array with a pair of values. You may compare numbers, strings, and dates. Except for $cmp (page 272), all comparison operators return a Boolean value. $cmp (page 272) returns an integer. $cmp Takes two values in an array and returns an integer. The returned value is: A negative number if the rst value is less than the second. A positive number if the rst value is greater than the second. 0 if the two values are equal. $eq Takes two values in an array and returns a boolean. The returned value is: true when the values are equivalent. false when the values are not equivalent. $gt Takes two values in an array and returns a boolean. The returned value is: true when the rst value is greater than the second value. false when the rst value is less than or equal to the second value. $gte Takes two values in an array and returns a boolean. The returned value is: true when the rst value is greater than or equal to the second value. false when the rst value is less than the second value. $lt Takes two values in an array and returns a boolean. The returned value is: true when the rst value is less than the second value. false when the rst value is greater than or equal to the second value. $lte Takes two values in an array and returns a boolean. The returned value is:
28.50. Expressions
303
true when the rst value is less than or equal to the second value. false when the rst value is greater than the second value. $ne Takes two values in an array returns a boolean. The returned value is: true when the values are not equivalent. false when the values are equivalent.
28.50.4 Arithmetic Operators

These operators only support numbers. $add Takes an array of one or more numbers and adds them together, returning the sum. $divide Takes an array that contains a pair of numbers and returns the value of the rst number divided by the second number. $mod Takes an array that contains a pair of numbers and returns the remainder of the rst number divided by the second number. See also: $mod (page 772) $multiply Takes an array of one or more numbers and multiples them, returning the resulting product. $subtract Takes an array that contains a pair of numbers and subtracts the second from the rst, returning their difference.
28.50.5 String Operators

These operators manipulate strings within projection expressions. $concat New in version 2.4. Takes an array of strings, concatenates the strings, and returns the concatenated string. $concat (page 272) can only accept an array of strings. Use $concat (page 272) with the following syntax:
{ $concat: [ <string>, <string>, ... ] }
If array element has a value of null or refers to a eld that is missing, $concat (page 272) will return null. $strcasecmp Takes in two strings. Returns a number. $strcasecmp (page 289) is positive if the rst string is greater than the second and negative if the rst string is less than the second. $strcasecmp (page 289) returns 0 if the strings are identical. Note: $strcasecmp (page 289) may not make sense when applied to glyphs outside the Roman alphabet. $strcasecmp (page 289) internally capitalizes strings before comparing them to provide a case-insensitive comparison. Use $cmp (page 272) for a case sensitive comparison.
304
$substr $substr (page 289) takes a string and two numbers. The rst number represents the number of bytes in the string to skip, and the second number species the number of bytes to return from the string. Note: $substr (page 289) is not encoding aware and if used improperly may produce a result string containing an invalid UTF-8 character sequence. $toLower Takes a single string and converts that string to lowercase, returning the result. All uppercase letters become lowercase. Note: $toLower (page 289) may not make sense when applied to glyphs outside the Roman alphabet. $toUpper Takes a single string and converts that string to uppercase, returning the result. All lowercase letters become uppercase. Note: $toUpper (page 290) may not make sense when applied to glyphs outside the Roman alphabet.
28.50.6 Date Operators

All date operators take a Date typed value as a single argument and return a number. $dayOfYear Takes a date and returns the day of the year as a number between 1 and 366. $dayOfMonth Takes a date and returns the day of the month as a number between 1 and 31. $dayOfWeek Takes a date and returns the day of the week as a number between 1 (Sunday) and 7 (Saturday.) $year Takes a date and returns the full year. $month Takes a date and returns the month as a number between 1 and 12. $week Takes a date and returns the week of the year as a number between 0 and 53. Weeks begin on Sundays, and week 1 begins with the rst Sunday of the year. Days preceding the rst Sunday of the year are in week 0. This behavior is the same as the %U operator to the strftime standard library function. $hour Takes a date and returns the hour between 0 and 23. $minute Takes a date and returns the minute between 0 and 59. $second Takes a date and returns the second between 0 and 59, but can be 60 to account for leap seconds. $millisecond Takes a date and returns the millisecond portion of the date as an integer between 0 and 999.
28.50. Expressions
305
28.50.7 Conditional Expressions

$cond Use the $cond (page 275) operator with the following syntax:
{ $cond: [ <boolean-expression>, <true-case>, <false-case> ] }
Takes an array with three expressions, where the rst expression evaluates to a Boolean value. If the rst expression evaluates to true, $cond (page 275) returns the value of the second expression. If the rst expression evaluates to false, $cond (page 275) evaluates and returns the third expression. $ifNull Use the $ifNull (page 280) operator with the following syntax:
{ $ifNull: [ <expression>, <replacement-if-null> ] }
Takes an array with two expressions. $ifNull (page 280) returns the rst expression if it evaluates to a non-null value. Otherwise, $ifNull (page 280) returns the second expressions value.
306
CHAPTER 29
SQL to Aggregation Framework Mapping Chart
The aggregation framework (page 255) allows MongoDB to provide native aggregation capabilities that corresponds to many common data aggregation operations in SQL. If youre new to MongoDB you might want to consider the Frequently Asked Questions (page 709) section for a selection of common questions. The following table provides an overview of common SQL aggregation terms, functions, and concepts and the corresponding MongoDB aggregation operators (page 291): SQL Terms, Functions, and Concepts WHERE GROUP BY HAVING SELECT ORDER BY LIMIT SUM() COUNT() join MongoDB Aggregation Operators
$match (page 281) $group (page 278) $match (page 281) $project (page 285) $sort (page 287) $limit (page 280) $sum $sum No direct corresponding operator; however, the $unwind (page 290) operator allows for somewhat similar functionality, but with elds embedded within the document.
29.1 Examples
The following table presents a quick reference of SQL aggregation statements and the corresponding MongoDB statements. The examples in the table assume the following conditions: The SQL examples assume two tables, orders and order_lineitem that join by the order_lineitem.order_id and the orders.id columns. The MongoDB examples assume one collection orders that contain documents of the following prototype:
{ cust_id: "abc123", ord_date: ISODate("2012-11-02T17:04:11.102Z"), status: A, price: 50, items: [ { sku: "xxx", qty: 25, price: 1 }, { sku: "yyy", qty: 25, price: 1 } ] }
307
The MongoDB statements prex the names of the elds from the documents in the collection orders with a $ character when they appear as operands to the aggregation operations.
308
Chapter 29. SQL to Aggregation Framework Mapping Chart
SQL Example SELECT COUNT(*) AS count FROM orders
MongoDB Example
db.orders.aggregate( [ { $group: { _id: null, count: { $sum: 1 } } } ] )
Description Count all records from orders
Sum the price eld from orders SELECT SUM(price) AS total db.orders.aggregate( [ FROM orders { $group: { _id: null, total: { $sum: "$price" } } } ] ) For each unique cust_id, sum the SELECT cust_id, db.orders.aggregate( [ price eld. SUM(price) AS total { $group: { _id: "$cust_id", FROM orders total: { $sum: "$price" } } } GROUP BY cust_id ] ) For each unique cust_id, sum the SELECT cust_id, db.orders.aggregate( [ price eld, results sorted by sum. SUM(price) AS total { $group: { _id: "$cust_id", FROM orders total: { $sum: "$price" } } }, GROUP BY cust_id { $sort: { total: 1 } } ORDER BY total ] ) For each unique cust_id, SELECT cust_id, db.orders.aggregate( [ ord_date grouping, sum the ord_date, { $group: { _id: { cust_id: "$cust_id", price eld. SUM(price) AS total ord_date: "$ord_date" }, FROM orders total: { $sum: "$price" } } } GROUP BY cust_id, ord_date ] ) SELECT cust_id, count(*) FROM orders GROUP BY cust_id HAVING count(*) > 1 For cust_id with multiple records, db.orders.aggregate( [ return the cust_id and the corre{ $group: { _id: "$cust_id", sponding record count. count: { $sum: 1 } } }, { $match: { count: { $gt: 1 } } } ] )
For each unique cust_id, SELECT cust_id, db.orders.aggregate( [ ord_date grouping, sum the ord_date, { $group: { _id: { cust_id: "$cust_id", price eld and return only where SUM(price) AS total ord_date: "$ord_date" }, the sum is greater than 250. FROM orders total: { $sum: "$price" } } }, GROUP BY cust_id, ord_date { $match: { total: { $gt: 250 } } } HAVING total > 250 ] ) For each unique cust_id with staSELECT cust_id, db.orders.aggregate( [ tus A, sum the price eld. SUM(price) as total { $match: { status: A } }, FROM orders { $group: { _id: "$cust_id", WHERE status = A total: { $sum: "$price" } } } GROUP BY cust_id ] ) For each unique cust_id with staSELECT cust_id, db.orders.aggregate( [ tus A, sum the price eld and return SUM(price) as total { $match: { status: A } }, only where the sum is greater than FROM orders { $group: { _id: "$cust_id", 250. WHERE status = A total: { $sum: "$price" } } }, 29.1. Examples 309 GROUP BY cust_id { $match: { total: { $gt: 250 } } } HAVING total > 250 ] ) For each unique cust_id, sum the
310
Chapter 29. SQL to Aggregation Framework Mapping Chart
CHAPTER 30
Map-Reduce
Map-reduce operations can handle complex aggregation tasks. To perform map-reduce operations, MongoDB provides the mapReduce (page 814) command and, in the mongo (page 1036) shell, the db.collection.mapReduce() (page 937) wrapper method.
30.1 Examples
For examples of map-reduce, see
30.1.1 Map-Reduce Examples

In the mongo (page 1036) shell, the db.collection.mapReduce() (page 937) method is a wrapper around the mapReduce (page 814) command. The following examples use the db.collection.mapReduce() (page 937) method: Consider the following map-reduce operations on a collection orders that contains documents of the following prototype:
{ _id: ObjectId("50a8240b927d5d8b5891743c"), cust_id: "abc123", ord_date: new Date("Oct 04, 2012"), status: A, price: 250, items: [ { sku: "mmm", qty: 5, price: 2.5 }, { sku: "nnn", qty: 5, price: 2.5 } ] }
Return the Total Price Per Customer Perform map-reduce operation on the orders collection to group by the cust_id, and for each cust_id, calculate the sum of the price for each cust_id:
1. Dene the map function to process each input document: In the function, this refers to the document that the map-reduce operation is processing.
311
The function maps the price to the cust_id for each document and emits the cust_id and price pair.
var mapFunction1 = function() { emit(this.cust_id, this.price); };
2. Dene the corresponding reduce function with two arguments keyCustId and valuesPrices: The valuesPrices is an array whose elements are the price values emitted by the map function and grouped by keyCustId. The function reduces the valuesPrice array to the sum of its elements.
var reduceFunction1 = function(keyCustId, valuesPrices) { return Array.sum(valuesPrices); };
3. Perform the map-reduce on all documents in the orders collection using the mapFunction1 map function and the reduceFunction1 reduce function.
db.orders.mapReduce( mapFunction1, reduceFunction1, { out: "map_reduce_example" } )
This operation outputs the results to a collection named map_reduce_example. If the map_reduce_example collection already exists, the operation will replace the contents with the results of this map-reduce operation: Calculate Order and Total Quantity with Average Quantity Per Item In this example you will perform a map-reduce operation on the orders collection, for all documents that have an ord_date value greater than 01/01/2012. The operation groups by the item.sku eld, and for each sku calculates the number of orders and the total quantity ordered. The operation concludes by calculating the average quantity per order for each sku value: 1. Dene the map function to process each input document: In the function, this refers to the document that the map-reduce operation is processing. For each item, the function associates the sku with a new object value that contains the count of 1 and the item qty for the order and emits the sku and value pair.
var mapFunction2 = function() { for (var idx = 0; idx < this.items.length; idx++) { var key = this.items[idx].sku; var value = { count: 1, qty: this.items[idx].qty }; emit(key, value); } };
2. Dene the corresponding reduce function with two arguments keySKU and countObjVals: countObjVals is an array whose elements are the objects mapped to the grouped keySKU values passed by map function to the reducer function.
312
Chapter 30. Map-Reduce
The function reduces the countObjVals array to a single object reducedValue that contains the count and the qty elds. In reducedVal, the count eld contains the sum of the count elds from the individual array elements, and the qty eld contains the sum of the qty elds from the individual array elements.
var reduceFunc2 = function(keySKU, countObjVals) { reducedVal = { count: 0, qty: 0 }; for (var idx = 0; idx < countObjVals.length; idx++) { reducedVal.count += countObjVals[idx].count; reducedVal.qty += countObjVals[idx].qty; } return reducedVal; };
3. Dene a nalize function with two arguments key and reducedVal. The function modies the reducedVal object to add a computed eld named avg and returns the modied object:
var finalizeFunc2 = function (key, reducedVal) { reducedVal.avg = reducedVal.qty/reducedVal.count; return reducedVal; };
4. Perform the map-reduce operation on the orders collection reduceFunction2, and finalizeFunction2 functions.
using
the
mapFunction2,
db.orders.mapReduce( mapFunc2, reduceFunc2, { out: { merge: "map_reduce_example" }, query: { ord_date: { $gt: new Date(01/01/2012) } }, finalize: finalizeFunc2 } )
This operation uses the query eld to select only those documents with ord_date greater than new Date(01/01/2012). Then it output the results to a collection map_reduce_example. If the map_reduce_example collection already exists, the operation will merge the existing contents with the results of this map-reduce operation:
30.1.2 Perform Incremental Map-Reduce

Map-reduce operations can handle complex aggregation tasks. To perform map-reduce operations, MongoDB provides the mapReduce (page 814) command and, in the mongo (page 1036) shell, the db.collection.mapReduce() (page 937) wrapper method. If the map-reduce dataset is constantly growing, then rather than performing the map-reduce operation over the entire dataset each time you want to run map-reduce, you may want to perform an incremental map-reduce. To perform incremental map-reduce: 1. Run a map-reduce job over the current collection and output the result to a separate collection. 30.1. Examples 313
2. When you have more data to process, run subsequent map-reduce job with: the query parameter that species conditions that match only the new documents. the out parameter that species the reduce action to merge the new results into the existing output collection. Consider the following example where you schedule a map-reduce operation on a sessions collection to run at the end of each day. Data Setup The sessions collection contains documents that log users session each day, for example:
db.sessions.save( db.sessions.save( db.sessions.save( db.sessions.save( db.sessions.save( db.sessions.save( db.sessions.save( db.sessions.save( { { { { { { { { userid: userid: userid: userid: userid: userid: userid: userid: "a", "b", "c", "d", "a", "b", "c", "d", ts: ts: ts: ts: ts: ts: ts: ts: ISODate(2011-11-03 ISODate(2011-11-03 ISODate(2011-11-03 ISODate(2011-11-03 ISODate(2011-11-04 ISODate(2011-11-04 ISODate(2011-11-04 ISODate(2011-11-04 14:17:00), 14:23:00), 15:02:00), 16:45:00), 11:05:00), 13:14:00), 17:00:00), 15:37:00), length: length: length: length: length: length: length: length: 95 } ); 110 } ); 120 } ); 45 } ); 105 } ); 120 } ); 130 } ); 65 } );
Initial Map-Reduce of Current Collection Run the rst map-reduce operation as follows: 1. Dene the map function that maps the userid to an object that contains the elds userid, total_time, count, and avg_time:
var mapFunction = function() { var key = this.userid; var value = { userid: this.userid, total_time: this.length, count: 1, avg_time: 0 }; emit( key, value ); };
2. Dene the corresponding reduce function with two arguments key and values to calculate the total time and the count. The key corresponds to the userid, and the values is an array whose elements corresponds to the individual objects mapped to the userid in the mapFunction.
var reduceFunction = function(key, values) { var reducedObject = { userid: key, total_time: 0, count:0, avg_time:0 }; values.forEach( function(value) { reducedObject.total_time += value.total_time;
314
reducedObject.count += value.count; } ); return reducedObject; };
3. Dene finalize function with two arguments key and reducedValue. The function modies the reducedValue document to add another eld average and returns the modied document.
var finalizeFunction = function (key, reducedValue) {
if (reducedValue.count > 0) reducedValue.avg_time = reducedValue.total_time / reducedValue.cou return reducedValue; };
4. Perform map-reduce on the session collection using the mapFunction, the reduceFunction, and the finalizeFunction functions. Output the results to a collection session_stat. If the session_stat collection already exists, the operation will replace the contents:
db.sessions.mapReduce( mapFunction, reduceFunction, { out: { reduce: "session_stat" }, finalize: finalizeFunction } )
Subsequent Incremental Map-Reduce Later as the sessions collection grows, you can run additional map-reduce operations. For example, add new documents to the sessions collection:
db.sessions.save( db.sessions.save( db.sessions.save( db.sessions.save( { { { { userid: userid: userid: userid: "a", "b", "c", "d", ts: ts: ts: ts: ISODate(2011-11-05 ISODate(2011-11-05 ISODate(2011-11-05 ISODate(2011-11-05 14:17:00), 14:23:00), 15:02:00), 16:45:00), length: length: length: length: 100 } ); 115 } ); 125 } ); 55 } );
At the end of the day, perform incremental map-reduce on the sessions collection but use the query eld to select only the new documents. Output the results to the collection session_stat, but reduce the contents with the results of the incremental map-reduce:
db.sessions.mapReduce( mapFunction, reduceFunction, { query: { ts: { $gt: ISODate(2011-11-05 00:00:00) } }, out: { reduce: "session_stat" }, finalize: finalizeFunction } );
For many simple aggregation tasks, see the aggregation framework (page 255).
30.1. Examples
315
30.2 Temporary Collection

The map-reduce operation uses a temporary collection during processing. At completion, the map-reduce operation renames the temporary collection. As a result, you can perform a map-reduce operation periodically with the same target collection name without affecting the intermediate states. Use this mode when generating statistical output collections on a regular basis.
30.3 Concurrency
The map-reduce operation is composed of many tasks, including: reads from the input collection, executions of the map function, executions of the reduce function, writes to the output collection. These various tasks take the following locks: The read phase takes a read lock. It yields every 100 documents. The insert into the temporary collection takes a write lock for a single write. If the output collection does not exist, the creation of the output collection takes a write lock. If the output collection exists, then the output actions (i.e. merge, replace, reduce) take a write lock. Changed in version 2.4: The V8 JavaScript engine, which became the default in 2.4, allows multiple JavaScript operations to execute at the same time. Prior to 2.4, JavaScript code (i.e. map, reduce, finalize functions) executed in a single thread. Note: The nal write lock during post-processing makes the results appear atomically. However, output actions merge and reduce may take minutes to process. For the merge and reduce, the nonAtomic ag is available. See the db.collection.mapReduce() (page 937) reference for more information.
30.4 Sharded Cluster

30.4.1 Sharded Input
When using sharded collection as the input for a map-reduce operation, mongos (page 1032) will automatically dispatch the map-reduce job to each shard in parallel. There is no special option required. mongos (page 1032) will wait for jobs on all shards to nish.
30.4.2 Sharded Output

By default the output collection is not sharded. The process is: mongos (page 1032) dispatches a map-reduce nish job to the shard that will store the target collection. The target shard pulls results from all other shards, and runs a nal reduce/nalize operation, and write to the output.
316
If using the sharded option to the out parameter, MongoDB shards the output using _id eld as the shard key. Changed in version 2.2. If the output collection does not exist, MongoDB creates and shards the collection on the _id eld. If the collection is empty, MongoDB creates chunks using the result of the rst stage of the map-reduce operation. mongos (page 1032) dispatches, in parallel, a map-reduce nish job to every shard that owns a chunk. Each shard will pull the results it owns from all other shards, run a nal reduce/nalize, and write to the output collection. Note: During later map-reduce jobs, MongoDB splits chunks as needed. Balancing of chunks for the output collection is automatically prevented during post-processing to avoid concurrency issues. In MongoDB 2.0: mongos (page 1032) retrieves the results from each shard, and performs merge sort to order the results, and performs a reduce/nalize as needed. mongos (page 1032) then writes the result to the output collection in sharded mode. This model requires only a small amount of memory, even for large datasets. Shard chunks are not automatically split during insertion. This requires manual intervention until the chunks are granular and balanced. Warning: For best results, only use the sharded output options for mapReduce (page 814) in version 2.2 or later.
30.5 Troubleshooting Map-Reduce Operations

You can troubleshoot the map function and the reduce function in the mongo (page 1036) shell. See the following tutorials for more information:
30.5.1 Troubleshoot the Map Function

The map function is a JavaScript function that associates or maps a value with a key and emits the key and value pair during a map-reduce (page 311) operation. To verify the key and value pairs emitted by the map function, write your own emit function. Consider a collection orders that contains documents of the following prototype:
30.5. Troubleshooting Map-Reduce Operations
317
1. Dene the map function that maps the price to the cust_id for each document and emits the cust_id and price pair:
var map = function() { emit(this.cust_id, this.price); };
2. Dene the emit function to print the key and value:

var emit = function(key, value) { print("emit"); print("key: " + key + " value: " + tojson(value)); }
3. Invoke the map function with a single document from the orders collection:
var myDoc = db.orders.findOne( { _id: ObjectId("50a8240b927d5d8b5891743c") } ); map.apply(myDoc);
4. Verify the key and value pair is as you expected.

emit key: abc123 value:250
5. Invoke the map function with multiple documents from the orders collection:
var myCursor = db.orders.find( { cust_id: "abc123" } ); while (myCursor.hasNext()) { var doc = myCursor.next(); print ("document _id= " + tojson(doc._id)); map.apply(doc); print(); }
6. Verify the key and value pairs are as you expected. See also: The map function must meet various requirements. For a list of all the requirements for the map function, see mapReduce (page 814), or the mongo (page 1036) shell helper method db.collection.mapReduce() (page 937).
30.5.2 Troubleshoot the Reduce Function

The reduce function is a JavaScript function that reduces to a single object all the values associated with a particular key during a map-reduce (page 311) operation. The reduce function must meet various requirements. This tutorial helps verify that the reduce function meets the following criteria: The reduce function must return an object whose type must be identical to the type of the value emitted by the map function. The order of the elements in the valuesArray should not affect the output of the reduce function. The reduce function must be idempotent. For a list of all the requirements for the reduce function, see mapReduce (page 814), or the mongo (page 1036) shell helper method db.collection.mapReduce() (page 937).
318
Conrm Output Type You can test that the reduce function returns a value that is the same type as the value emitted from the map function. 1. Dene a reduceFunction1 function that takes the arguments keyCustId and valuesPrices. valuesPrices is an array of integers:
2. Dene a sample array of integers:

var myTestValues = [ 5, 5, 10 ];
3. Invoke the reduceFunction1 with myTestValues:

reduceFunction1(myKey, myTestValues);
4. Verify the reduceFunction1 returned an integer:

20
5. Dene a reduceFunction2 function that takes the arguments keySKU and valuesCountObjects. valuesCountObjects is an array of documents that contain two elds count and qty:
var reduceFunction2 = function(keySKU, valuesCountObjects) { reducedValue = { count: 0, qty: 0 }; for (var idx = 0; idx < valuesCountObjects.length; idx++) { reducedValue.count += valuesCountObjects[idx].count; reducedValue.qty += valuesCountObjects[idx].qty; } return reducedValue; };
6. Dene a sample array of documents:

var myTestObjects = [ { count: 1, qty: 5 }, { count: 2, qty: 10 }, { count: 3, qty: 15 } ];
7. Invoke the reduceFunction2 with myTestObjects:

reduceFunction2(myKey, myTestObjects);
8. Verify the reduceFunction2 returned a document with exactly the count and the qty eld:
{ "count" : 6, "qty" : 30 }
Ensure Insensitivity to the Order of Mapped Values The reduce function takes a key and a values array as its argument. You can test that the result of the reduce function does not depend on the order of the elements in the values array. 1. Dene a sample values1 array and a sample values2 array that only differ in the order of the array elements:
319
var values1 = [ { count: 1, qty: 5 }, { count: 2, qty: 10 }, { count: 3, qty: 15 } ]; var values2 = [ { count: 3, qty: 15 }, { count: 1, qty: 5 }, { count: 2, qty: 10 } ];
2. Dene a reduceFunction2 function that takes the arguments keySKU and valuesCountObjects. valuesCountObjects is an array of documents that contain two elds count and qty:
3. Invoke the reduceFunction2 rst with values1 and then with values2:
reduceFunction2(myKey, values1); reduceFunction2(myKey, values2);
4. Verify the reduceFunction2 returned the same result:

{ "count" : 6, "qty" : 30 }
Ensure Reduce Function Idempotence Because the map-reduce operation may call a reduce multiple times for the same key, the reduce function must return a value of the same type as the value emitted from the map function. You can test that the reduce function process reduced values without affecting the nal value. 1. Dene a reduceFunction2 function that takes the arguments keySKU and valuesCountObjects. valuesCountObjects is an array of documents that contain two elds count and qty:
2. Dene a sample key:
320
var myKey = myKey;
3. Dene a sample valuesIdempotent array that contains an element that is a call to the reduceFunction2 function:
var valuesIdempotent = [ { count: 1, qty: 5 }, { count: 2, qty: 10 }, reduceFunction2(myKey, [ { count:3, qty: 15 } ] ) ];
4. Dene a sample values1 array that combines the values passed to reduceFunction2:
var values1 = [ { count: 1, qty: 5 }, { count: 2, qty: 10 }, { count: 3, qty: 15 } ];
5. Invoke the reduceFunction2 rst with myKey and valuesIdempotent and then with myKey and values1:
reduceFunction2(myKey, valuesIdempotent); reduceFunction2(myKey, values1);
6. Verify the reduceFunction2 returned the same result:

{ "count" : 6, "qty" : 30 }
321
322
CHAPTER 31
Simple Aggregation Methods and Commands
In addition to the aggregation framework (page 255) and map-reduce, MongoDB provides the following methods and commands to perform aggregation:
31.1 Count
MongoDB offers the following command and methods to provide count functionality: count (page 808) db.collection.count() (page 919) cursor.count() (page 952)
31.2 Distinct
MongoDB offers the following command and method to provide the distinct functionality: distinct (page 809) db.collection.distinct() (page 920)
31.3 Group
MongoDB offers the following command and method to provide group functionality: group (page 810) db.collection.group() (page 932)
323
324
Chapter 31. Simple Aggregation Methods and Commands
Part VII
Indexes
325
Indexes provide high performance read operations for frequently used queries. Indexes are particularly useful where the total size of the documents exceeds the amount of available RAM. For basic concepts and options, see Indexing Overview (page 329). For procedures and operational concerns, see Indexing Operations (page 347). For information on how applications might use indexes, see Indexing Strategies (page 341).
327
328
CHAPTER 32
Index Concepts
32.1 Indexing Overview

This document provides an overview of indexes in MongoDB, including index types and creation options. For operational guidelines and procedures, see the Indexing Operations (page 347) document. For strategies and practical approaches, see the Indexing Strategies (page 341) document.
32.1.1 Synopsis
An index is a data structure that allows you to quickly locate documents based on the values stored in certain specied elds. Fundamentally, indexes in MongoDB are similar to indexes in other database systems. MongoDB supports indexes on any eld or sub-eld contained in documents within a MongoDB collection. MongoDB indexes have the following core features: MongoDB denes indexes on a per-collection level. You can create indexes on a single eld or on multiple elds using a compound index (page 331). Indexes enhance query performance, often dramatically. However, each index also incurs some overhead for every write operation. Consider the queries, the frequency of these queries, the size of your working set, the insert load, and your applications requirements as you create indexes in your MongoDB environment. All MongoDB indexes use a B-tree data structure. MongoDB can use this representation of the data to optimize query responses. Every query, including update operations, uses one and only one index. The query optimizer (page 174) selects the index empirically by occasionally running alternate query plans and by selecting the plan with the best response time for each query type. You can override the query optimizer using the cursor.hint() (page 958) method. An index covers a query if: all the elds in the query (page 168) are part of that index, and all the elds returned in the documents that match the query are in the same index. When an index covers a query, the server can both match the query conditions (page 168) and return the results using only the index; MongoDB does not need to look at the documents, only the index, to fulll the query. Querying the index can be faster than querying the documents outside of the index. See Create Indexes that Support Covered Queries (page 342) for more information. Using queries with good index coverage reduces the number of full documents that MongoDB needs to store in memory, thus maximizing database performance and throughput. 329
If an update does not change the size of a document or cause the document to outgrow its allocated area, then MongoDB will update an index only if the indexed elds have changed. This improves performance. Note that if the document has grown and must move, all index keys must then update.
32.1.2 Index Types

This section enumerates the types of indexes available in MongoDB. For all collections, MongoDB creates the default _id index (page 330). You can create additional indexes with the ensureIndex() (page 921) method on any single eld or sequence of elds (page 331) within any document or sub-document (page 330). MongoDB also supports indexes of arrays, called multi-key indexes (page 332). _id Index The _id index is a unique index (page 334) 1 on the _id eld, and MongoDB creates this index by default on all collections. 2 You cannot delete the index on _id. The _id eld is the primary key for the collection, and every document must have a unique _id eld. You may store any unique value in the _id eld. The default value of _id is an ObjectID on every insert() (page 936) operation. An ObjectId is a 12-byte unique identiers suitable for use as the value of an _id eld. Note: In sharded clusters, if you do not use the _id eld as the shard key, then your application must ensure the uniqueness of the values in the _id eld to prevent errors. This is most-often done by using a standard auto-generated ObjectId.
Secondary Indexes All indexes in MongoDB are secondary indexes. You can create indexes on any eld within any document or subdocument. Additionally, you can create compound indexes with multiple elds, so that a single query can match multiple components using the index while scanning fewer whole documents. In general, you should create indexes that support your primary, common, and user-facing queries. Doing so requires MongoDB to scan the fewest number of documents possible. In the mongo (page 1036) shell, you can create an index by calling the ensureIndex() (page 921) method. Arguments to ensureIndex() (page 921) resemble the following:
{ "field": 1 } { "product.quantity": 1 } { "product": 1, "quantity": 1 }
For each eld in the index specify either 1 for an ascending order or -1 for a descending order, which represents the order of the keys in the index. For indexes with more than one key (i.e. compound indexes (page 331)) the sequence of elds is important. Indexes on Sub-documents You can create indexes on elds that hold sub-documents as in the following example: Example
1 Although the index on _id is unique, the getIndexes() (page 930) method will not print unique: true in the mongo (page 1036) shell. 2 Before version 2.2 capped collections did not have an _id eld. In 2.2, all capped collections have an _id eld, except those in the local database. See the release notes (page 1162) for more information.
330
Chapter 32. Index Concepts
Given the following document in the factories collection:

{ "_id": ObjectId(...), metro: { city: "New York", state: "NY" } } )
You can create an index on the metro key. The following queries would then use that index, and both would return the above document:
db.factories.find( { metro: { city: "New York", state: "NY" } } ); db.factories.find( { metro: { $gte : { city: "New York" } } } );
The second query returns the document because { city: "New York" } is less than { city: "New York", state: "NY" } The order of comparison is in ascending key order in the order the keys occur in the BSON document.
Indexes on Embedded Fields You can create indexes on elds in sub-documents, just as you can index top-level elds in documents. indexes allow you to use a dot notation, to introspect into sub-documents. Consider a collection named people that holds documents that resemble the following example document:
{"_id": ObjectId(...) "name": "John Doe" "address": { "street": "Main" "zipcode": 53511 "state": "WI" } }
3
These
You can create an index on the address.zipcode eld, using the following specication:
db.people.ensureIndex( { "address.zipcode": 1 } )
Compound Indexes MongoDB supports compound indexes, where a single index structure holds references to multiple elds within a collections documents. Consider a collection named products that holds documents that resemble the following document:
{ "_id": ObjectId(...) "item": "Banana" "category": ["food", "produce", "grocery"] "location": "4th Street Store" "stock": 4 "type": cases "arrival": Date(...) }
If most applications queries include the item eld and a signicant number of queries will also check the stock eld, you can specify a single compound index to support both of these queries:
3 Indexes on Sub-documents (page 330), by contrast allow you to index elds that hold documents, including the full content, up to the maximum Index Size (page 1106) of the sub-document in the index.
32.1. Indexing Overview
331
db.products.ensureIndex( { "item": 1, "location": 1, "stock": 1 } )
Compound indexes support queries on any prex of the elds in the index. 4 For example, MongoDB can use the above index to support queries that select the item eld and to support queries that select the item eld and the location eld. The index, however, would not support queries that select the following: only the location eld only the stock eld only the location and stock elds only the item and stock elds Important: You may not create compound indexes that have hashed index elds. You will receive an error if you attempt to create a compound index that includes a hashed index (page 350). When creating an index, the number associated with a key species the direction of the index. The options are 1 (ascending) and -1 (descending). Direction doesnt matter for single key indexes or for random access retrieval but is important if you are doing sort queries on compound indexes. The order of elds in a compound index is very important. In the previous example, the index will contain references to documents sorted rst by the values of the item eld and, within each value of the item eld, sorted by the values of location, and then sorted by values of the stock eld. Indexes with Ascending and Descending Keys Indexes store references to elds in either ascending or descending order. For single-eld indexes, the order of keys doesnt matter, because MongoDB can traverse the index in either direction. However, for compound indexes (page 331), if you need to order results against two elds, sometimes you need the index elds running in opposite order relative to each other. To specify an index with a descending order, use the following form:
db.products.ensureIndex( { "field": -1 } )
More typically in the context of a compound index (page 331), the specication would resemble the following prototype:
db.products.ensureIndex( { "fieldA": 1, "fieldB": -1 } )
Consider a collection of event data that includes both usernames and a timestamp. If you want to return a list of events sorted by username and then with the most recent events rst. To create this index, use the following command:
db.events.ensureIndex( { "username" : 1, "timestamp" : -1 } )
Multikey Indexes If you index a eld that contains an array, MongoDB indexes each value in the array separately, in a multikey index. Example Given the following document:
4
a:
Index prexes are the beginning subset of elds. For example, given the index { a: 1, b: 1 } are prexes of the index.
1, b:
1, c:
1 } both { a:
1 } and {
332
{ "_id" : ObjectId("..."), "name" : "Warm Weather", "author" : "Steve", "tags" : [ "weather", "hot", "record", "april" ] }
Then an index on the tags eld would be a multikey index and would include these separate entries:
{ { { { tags: tags: tags: tags: "weather" } "hot" } "record" } "april" }
Queries could use the multikey index to return queries for any of the above values. Note: For hashed indexes, MongoDB collapses sub-documents and computes the hash for the entire value, but does not support multi-key (i.e. arrays) indexes. For elds that hold sub-documents, you cannot use the index to support queries that introspect the sub-document. You can use multikey indexes to index elds within objects embedded in arrays, as in the following example: Example Consider a feedback collection with documents in the following form:
{ "_id": ObjectId(...) "title": "Grocery Quality" "comments": [ { author_id: ObjectId(...) date: Date(...) text: "Please expand the cheddar selection." }, { author_id: ObjectId(...) date: Date(...) text: "Please expand the mustard selection." }, { author_id: ObjectId(...) date: Date(...) text: "Please expand the olive selection." } ] }
An index on the comments.text eld would be a multikey index and would add items to the index for all of the sub-documents in the array. With an index, such as { comments.text: 1 }, consider the following query:
db.feedback.find( { "comments.text": "Please expand the selection." } )
This would select the document, that contains the following document in the comments.text array:
{ author_id: ObjectId(...) date: Date(...) text: "Please expand the olive selection." }
Compound Multikey Indexes May Only Include One Array Field While you can create multikey compound indexes (page 331), at most one eld in a compound index may hold an array. For example, given an index on { a: 1, b: 1 }, the following documents are permissible:
333
{a: [1, 2], b: 1} {a: 1, b: [1, 2]}
However, the following document is impermissible, and MongoDB cannot insert such a document into a collection with the {a: 1, b: 1 } index:
{a: [1, 2], b: [1, 2]}
If you attempt to insert a such a document, MongoDB will reject the insertion, and produce an error that says cannot index parallel arrays. MongoDB does not index parallel arrays because they require the index to include each value in the Cartesian product of the compound keys, which could quickly result in incredibly large and difcult to maintain indexes.
Unique Indexes A unique index causes MongoDB to reject all documents that contain a duplicate value for the indexed eld. To create a unique index on the user_id eld of the members collection, use the following operation in the mongo (page 1036) shell:
db.addresses.ensureIndex( { "user_id": 1 }, { unique: true } )
By default, unique is false on MongoDB indexes. If you use the unique constraint on a compound index (page 331) then MongoDB will enforce uniqueness on the combination of values, rather than the individual value for any or all values of the key. If a document does not have a value for the indexed eld in a unique index, the index will store a null value for this document. MongoDB will only permit one document without a unique value in the collection because of this unique constraint. You can combine with the sparse index (page 334) to lter these null values from the unique index. You may not specify a unique constraint on a hashed index (page 335). Sparse Indexes Sparse indexes only contain entries for documents that have the indexed eld. 5 Any document that is missing the eld is not indexed. The index is sparse because of the missing documents when values are missing. By contrast, non-sparse indexes contain all documents in a collection, and store null values for documents that do not contain the indexed eld. Create a sparse index on the xmpp_id eld, of the members collection, using the following operation in the mongo (page 1036) shell:
db.addresses.ensureIndex( { "xmpp_id": 1 }, { sparse: true } )
By default, sparse is false on MongoDB indexes. Warning: Using these indexes will sometimes result in incomplete results when ltering or sorting results, because sparse indexes are not complete for all documents in a collection.
Note: Do not confuse sparse indexes in MongoDB with block-level indexes in other databases. Think of them as dense indexes with a specic lter. You can combine the sparse index option with the unique indexes (page 334) option so that mongod (page 1021) will reject documents that have duplicate values for a eld, but that ignore documents that do not have the key.
5
All documents that have the indexed eld are indexed in a sparse index, even if that eld stores a null value in some documents.
334
Hashed Index New in version 2.4. Hashed indexes maintain entries with hashes of the values of the indexed eld. The hashing function collapses subdocuments and computes the hash for the entire value but does not support multi-key (i.e. arrays) indexes. MongoDB can use the hashed index to support equality queries, but hashed indexes do not support range queries. You may not create compound indexes that have hashed index elds or specify a unique constraint on a hashed index; however, you can create both a hashed index and an ascending/descending (i.e. non-hashed) index on the same eld: MongoDB will use the scalar index for range queries. Warning: hashed indexes truncate oating point numbers to 64-bit integers before hashing. For example, a hashed index would store the same value for a eld that held a value of 2.3, 2.2 and 2.9. To prevent collisions, do not use a hashed index for oating point numbers that cannot be consistently converted to 64-bit integers (and then back to oating point.) hashed indexes do not support oating point values larger than 253 . Create a hashed index using an operation that resembles the following:
db.active.ensureIndex( { a: "hashed" } )
This operation creates a hashed index for the active collection on the a eld.
32.1.3 Index Names

The default name for an index is the concatenation of the indexed keys and each keys direction in the index (1 or -1). Example Issue the following command to create an index on item and quantity:
db.products.ensureIndex( { item: 1, quantity: -1 } )
The resulting index is named: item_1_quantity_-1. Optionally, you can specify a name for an index instead of using the default name. Example Issue the following command to create an index on item and quantity and specify inventory as the index name:
db.products.ensureIndex( { item: 1, quantity: -1 } , {name: "inventory"} )
The resulting index is named: inventory. To view the name of an index, use the getIndexes() (page 930) method.
32.1.4 Index Creation Options

You specify index creation options in the second argument in ensureIndex() (page 921).
335
The options sparse (page 334), unique (page 334),and TTL (page 337) affect the kind of index that MongoDB creates. This section addresses, background construction (page 336) and duplicate dropping (page 337), which affect how MongoDB builds the indexes. Background Construction By default, creating an index is a blocking operation. Building an index on a large collection of data can take a long time to complete. To resolve this issue, the background option can allow you to continue to use your mongod (page 1021) instance during the index build. For example, to create an index in the background of the zipcode eld of the people collection you would issue the following:
db.people.ensureIndex( { zipcode: 1}, {background: true} )
By default, background is false for building MongoDB indexes. You can combine the background option with other options, as in the following:
db.people.ensureIndex( { zipcode: 1}, {background: true, sparse: true } )
Be aware of the following behaviors with background index construction: A mongod (page 1021) instance can build more than one index in the background concurrently. Changed in version 2.4: Before 2.4, a mongod (page 1021) instance could only build one background index per database at a time. Changed in version 2.2: Before 2.2, a single mongod (page 1021) instance could only build one index at a time. The indexing operation runs in the background so that other database operations can run while creating the index. However, the mongo (page 1036) shell session or connection where you are creating the index will block until the index build is complete. Open another connection or mongo (page 1036) instance to continue using commands to the database. The background index operation use an incremental approach that is slower than the normal foreground index builds. If the index is larger than the available RAM, then the incremental process can take much longer than the foreground build. If your application includes ensureIndex() (page 921) operations, and an index doesnt exist for other operational concerns, building the index can have a severe impact on the performance of the database. Make sure that your application checks for the indexes at start up using the getIndexes() (page 930) method or the equivalent method for your driver and terminates if the proper indexes do not exist. Always build indexes in production instances using separate application code, during designated maintenance windows. Building Indexes on Secondaries Background index operations on a replica set primary become foreground indexing operations on secondary members of the set. All indexing operations on secondaries block replication. To build large indexes on secondaries the best approach is to restart one secondary at a time in standalone mode and build the index. After building the index, restart as a member of the replica set, allow it to catch up with the other members of the set, and then build the index on the next secondary. When all the secondaries have the new index, step down the primary, restart it as a standalone, and build the index on the former primary. Remember, the amount of time required to build the index on a secondary node must be within the window of the oplog, so that the secondary can catch up with the primary. See Build Indexes on Replica Sets (page 351) for more information on this process.
336
Indexes on secondary members in recovering mode are always built in the foreground to allow them to catch up as soon as possible. See Build Indexes on Replica Sets (page 351) for a complete procedure for rebuilding indexes on secondaries. Note: If MongoDB is building an index in the background, you cannot perform other administrative operations involving that collection, including repairDatabase (page 868), drop that collection (i.e. db.collection.drop() (page 921),) and compact (page 859). These operations will return an error during background index builds. Queries will not use these indexes until the index build is complete. Drop Duplicates MongoDB cannot create a unique index (page 334) on a eld that has duplicate values. To force the creation of a unique index, you can specify the dropDups option, which will only index the rst occurrence of a value for the key, and delete all subsequent values. Warning: As in all unique indexes, if a document does not have the indexed eld, MongoDB will include it in the index with a null value. If subsequent elds do not have the indexed eld, and you have set {dropDups: true}, MongoDB will remove these documents from the collection when creating the index. If you combine dropDups with the sparse (page 334) option, this index will only include documents in the index that have the value, and the documents without the eld will remain in the database. To create a unique index that drops duplicates on the username eld of the accounts collection, use a command in the following form:
db.accounts.ensureIndex( { username: 1 }, { unique: true, dropDups: true } )
Warning: Specifying { dropDups: tion. By default, dropDups is false.
true } will delete data from your database. Use with extreme cau-
32.1.5 Index Features

TTL Indexes TTL indexes are special indexes that MongoDB can use to automatically remove documents from a collection after a certain amount of time. This is ideal for some types of information like machine generated event data, logs, and session information that only need to persist in a database for a limited amount of time. These indexes have the following limitations: Compound indexes (page 331) are not supported. The indexed eld must be a date type. If the eld holds an array, and there are multiple date-typed data in the index, the document will expire when the lowest (i.e. earliest) matches the expiration threshold. Note: TTL indexes expire data by removing documents in a background task that runs every 60 seconds. As a result, the TTL index provides no guarantees that expired documents will not exist in the collection. Consider that: Documents may remain in a collection after they expire and before the background process runs. 32.1. Indexing Overview 337
The duration of the removal operations depend on the workload of your mongod (page 1021) instance. In all other respects, TTL indexes are normal indexes, and if appropriate, MongoDB can use these indexes to fulll arbitrary queries. See also: Expire Data from Collections by Setting TTL (page 577) Geospatial Indexes MongoDB provides geospatial indexes to support location-based and other similar queries in a two dimensional coordinate systems. For example, use geospatial indexes when you need to take a collection of documents that have coordinates, and return a number of options that are near a given coordinate pair. To create a geospatial index, your documents must have a coordinate pair. For maximum compatibility, these coordinate pairs should be in the form of a two element array, such as [ x , y ]. Given the eld of loc, that held a coordinate pair, in the collection places, you would create a geospatial index as follows:
db.places.ensureIndex( { loc : "2d" } )
MongoDB will reject documents that have values in the loc eld beyond the minimum and maximum values. Note: MongoDB permits only one geospatial index per collection. Although, MongoDB will allow clients to create multiple geospatial indexes, a single query can use only one index. See the $near (page 780), and the database command geoNear (page 822) for more information on accessing geospatial data. Geohaystack Indexes In addition to conventional geospatial indexes (page 338), MongoDB also provides a bucket-based geospatial index, called geospatial haystack indexes. These indexes support high performance queries for locations within a small area, when the query must lter along another dimension. Example If you need to return all documents that have coordinates within 25 miles of a given point and have a type eld value of museum, a haystack index would be provide the best support for these queries. Haystack indexes allow you to tune your bucket size to the distribution of your data, so that in general you search only very small regions of 2d space for a particular kind of document. These indexes are not suited for nding the closest documents to a particular location, when the closest documents are far away compared to bucket size. text Indexes New in version 2.4. MongoDB provides text indexes to support the search of string content in documents of a collection. text indexes are case-insensitive and can include any eld that contains string data. text indexes drop language-specic stop words (e.g. in English, the, an, a, and, etc.) and uses simple language-specic sufx stemming. See Text Search Languages (page 836) for the supported languages. You can only access the text index with the text (page 832) command. See Text Search (page 371) for more information. 338 Chapter 32. Index Concepts
32.1.6 Index Behaviors

Limitations A collection may have no more than 64 indexes (page 1106). Index keys can be no larger than 1024 bytes (page 1106). Documents with elds that have values greater than this size cannot be indexed. To query for documents that were too large to index, you can use a command similar to the following:
db.records.find({<key>: <value too large to index>}).hint({$natural: 1})
The name of an index, including the namespace must be shorter than 128 characters (page 1106). Indexes have storage requirements, and impacts insert/update speed to some degree. Create indexes to support queries and other operations, but do not maintain indexes that your MongoDB instance cannot or will not use. For queries with the $or (page 770) operator, each clause of an $or (page 770) query executes in parallel, and can each use a different index. For queries that use the sort() (page 965) method and use the $or (page 770) operator, the query cannot use the indexes on the $or (page 770) elds. 2d geospatial queries (page 369) do not support queries that use the $or (page 770) operator. Consider Insert Throughput If your application is write-heavy, then be careful when creating new indexes, since each additional index with impose a write-performance penalty. In general, dont be careless about adding indexes. Add indexes to complement your queries. Always have a good reason for adding a new index, and be sure to benchmark alternative strategies. MongoDB must update all indexes associated with a collection after every insert, update, or delete operation. For update operations, if the updated document does not move to a new location, then MongoDB only modies the updated elds in the index. Therefore, every index on a collection adds some amount of overhead to these write operations. In almost every case, the performance gains that indexes realize for read operations are worth the insertion penalty. However, in some cases: An index to support an infrequent query might incur more insert-related costs than savings in read-time. If you have many related indexes on a collection that receives a high volume of write operations, you may nd better overall performance with a smaller number of indexes, even if some queries are less optimally supported by an index. If your indexes and queries are not sufciently selective (page 345), the speed improvements for query operations may not offset the costs of maintaining an index. For more information see Create Queries that Ensure Selectivity (page 345).
339
340
CHAPTER 33
Indexing Strategies for Applications
33.1 Indexing Strategies

The best indexes for your application are based on a number of factors, including the kinds of queries you expect, the ratio of reads to writes, and the amount of free memory on your system. When developing your indexing strategy you should have a deep understanding of: The applications queries. The relative frequency of each query in the application. The current indexes created for your collections. Which indexes the most common queries use. The best overall strategy for designing indexes is to prole a variety of index congurations with data sets similar to the ones youll be running in production to see which congurations perform best. MongoDB can only use one index to support any given operation. However, each clause of an $or (page 770) query may use a different index. The following documents introduce indexing strategies:
33.1.1 Create Indexes to Support Your Queries

An index supports a query when the index contains all the elds scanned by the query. The query scans the index and not the collection. Creating indexes that supports queries results in greatly increased query performance. This document describes strategies for creating indexes that support queries. Create a Single-Key Index if All Queries Use the Same, Single Key If you only ever query on a single key in a given collection, then you need to create just one single-key index for that collection. For example, you might create an index on category in the product collection:
db.products.ensureIndex( { "category": 1 } )
341
Create Compound Indexes to Support Several Different Queries If you sometimes query on only one key and at other times query on that key combined with a second key, then creating a compound index is more efcient than creating a single-key index. MongoDB will use the compound index for both queries. For example, you might create an index on both category and item.
db.products.ensureIndex( { "category": 1, "item": 1 } )
This allows you both options. You can query on just category, and you also can query on category combined with item. A single compound index (page 331) on multiple elds can support all the queries that search a prex subset of those elds. Note: With the exception of queries that use the $or (page 770) operator, a query does not use multiple indexes. A query uses only one index. Example The following index on a collection:
{ x: 1, y: 1, z: 1 }
Can support queries that the following indexes support:

{ x: 1 } { x: 1, y: 1 }
There are some situations where the prex indexes may offer better query performance: for example if z is a large array. The { x: 1, y: 1, z: 1 } index can also support many of the same queries as the following index:
{ x: 1, z: 1 }
Also, { x:
1, z:
1 } has an additional use. Given the following query:
db.collection.find( { x: 5 } ).sort( { z: 1} )
The { x: 1, z: 1 } index supports both the query and the sort operation, while the { x: 1, y: 1, z: 1 } index only supports the query. For more information on sorting, see Use Indexes to Sort Query Results (page 343).
Create Indexes that Support Covered Queries A covered query is a query in which: all the elds in the query (page 168) are part of an index, and all the elds returned in the results are in the same index. Because the index covers the query, MongoDB can both match the query conditions (page 168) and return the results using only the index; MongoDB does not need to look at the documents, only the index, to fulll the query. Querying only the index can be much faster than querying documents outside of the index. Index keys are typically smaller than the documents they catalog, and indexes are typically available in RAM or located sequentially on disk. MongoDB automatically uses an index that covers a query when possible. To ensure that an index can cover a query, create an index that includes all the elds listed in the query document (page 168) and in the query result. You can specify the elds to return in the query results with a projection (page 171) document. By default, MongoDB includes
342
Chapter 33. Indexing Strategies for Applications
the _id eld in the query result. So, if the index does not include the _id eld, then you must exclude the _id eld (i.e. _id: 0) from the query results. Example Given collection users with an index on the elds user and status, as created by the following option:
db.users.ensureIndex( { status: 1, user: 1 } )
Then, this index will cover the following query which selects on the status eld and returns only the user eld:
db.users.find( { status: "A" }, { user: 1, _id: 0 } )
In the operation, the projection document explicitly species _id: the index is only on the status and the user elds.
0 to exclude the _id eld from the result since
If the projection document does not specify the exclusion of the _id eld, the query returns the _id eld. The following query is not covered by the index on the status and the user elds because with the projection document { user: 1 }, the query returns both the user eld and the _id eld:
db.users.find( { status: "A" }, { user: 1 } )
An index cannot cover a query if: any of the indexed elds in any of the documents in the collection includes an array. If an indexed eld is an array, the index becomes a multi-key index (page 332) index and cannot support a covered query. any of the indexed elds are elds in subdocuments. To index elds in subdocuments, use dot notation. For example, consider a collection users with documents of the following form:
{ _id: 1, user: { login: "tester" } }
The collection has the following indexes:

{ user: 1 } { "user.login": 1 }
The { user:
1 } index covers the following query:
db.users.find( { user: { login: "tester" } }, { user: 1, _id: 0 } )
However, the { "user.login":
1 } index does not cover the following query:
db.users.find( { "user.login": "tester" }, { "user.login": 1, _id: 0 } )
The query, however, does use the { "user.login":
1 } index to nd matching documents.
To determine whether a query is a covered query, use the explain() (page 953) method. If the explain() (page 953) output displays true for the indexOnly (page 956) eld, the query is covered by an index, and MongoDB queries only that index to match the query and return the results. For more information see Measure Index Use (page 355).
33.1.2 Use Indexes to Sort Query Results

For the fastest performance when sorting query results by a given eld, create a sorted index on that eld. To sort query results on multiple elds, create a compound index (page 331). MongoDB sorts results based on the eld order in the index. For queries that include a sort that uses a compound index, ensure that all elds before the rst sorted eld are equality matches. 33.1. Indexing Strategies 343
Example If you create the following index:

{ a: 1, b: 1, c: 1, d: 1 }
The following query and sort operations can use the index:
db.collection.find().sort( { a:1 } ) db.collection.find().sort( { a:1, b:1 } ) db.collection.find( { a:4 } ).sort( { a:1, b:1 } ) db.collection.find( { b:5 } ).sort( { a:1, b:1 } ) db.collection.find( { a:5 } ).sort( { b:1, c:1 } ) db.collection.find( { a:5, c:4, b:3 } ).sort( { d:1 } ) db.collection.find( { a: { $gt:4 } } ).sort( { a:1, b:1 } ) db.collection.find( { a: { $gt:5 } } ).sort( { a:1, b:1 } ) db.collection.find( { a:5, b:3, d:{ $gt:4 } } ).sort( { c:1 } ) db.collection.find( { a:5, b:3, c:{ $lt:2 }, d:{ $gt:4 } } ).sort( { c:1 } )
However, the following queries cannot sort the results using the index:
db.collection.find().sort( { b:1 } ) db.collection.find( { b:5 } ).sort( { b:1 } )
Note: For in-memory sorts that do not use an index, the sort() (page 965) operation is signicantly slower. The sort() (page 965) operation will abort when it uses 32 megabytes of memory.
33.1.3 Ensure Indexes Fit RAM

For the fastest processing, ensure that your indexes t entirely in RAM so that the system can avoid reading the index from disk. To check the size of your indexes, use the db.collection.totalIndexSize() helper, which returns data in bytes:
> db.collection.totalIndexSize() 4294976499
The above example shows an index size of almost 4.3 gigabytes. To ensure this index ts in RAM, you must not only have more than that much RAM available but also must have RAM available for the rest of the working set. Also remember: If you have and use multiple collections, you must consider the size of all indexes on all collections. The indexes and the working set must be able to t in memory at the same time. There are some limited cases where indexes do not need to t in memory. See Indexes that Hold Only Recent Values in RAM (page 345). See also: collStats (page 873) and db.collection.stats() (page 947)
344
Indexes that Hold Only Recent Values in RAM Indexes do not have to t entirely into RAM in all cases. If the value of the indexed eld increments with every insert, and most queries select recently added documents; then MongoDB only needs to keep the parts of the index that hold the most recent or right-most values in RAM. This allows for efcient index use for read and write operations and minimize the amount of RAM required to support the index.
33.1.4 Create Queries that Ensure Selectivity

Selectivity is the ability of a query to narrow results using the index. Effective indexes are more selective and allow MongoDB to use the index for a larger portion of the work associated with fullling the query. To ensure selectivity, write queries that limit the number of possible documents with the indexed eld. Write queries that are appropriately selective relative to your indexed data. Example Suppose you have a eld called status where the possible values are new and processed. If you add an index on status youve created a low-selectivity index. The index will be of little help in locating records. A better strategy, depending on your queries, would be to create a compound index (page 331) that includes the lowselectivity eld and another eld. For example, you could create a compound index on status and created_at. Another option, again depending on your use case, might be to use separate collections, one for each status. Example Consider an index { a : 1 } (i.e. an index on the key a sorted in ascending order) on a collection where a has three values evenly distributed across the collection:
{ { { { { { { { { _id: _id: _id: _id: _id: _id: _id: _id: _id: ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), a: a: a: a: a: a: a: a: a: 1, 1, 1, 2, 2, 2, 3, 3, 3, b: b: b: b: b: b: b: b: b: "ab" "cd" "ef" "jk" "lm" "no" "pq" "rs" "tv" } } } } } } } } }
If you query for { a: 2, b: "no" } MongoDB must scan 3 documents in the collection to return the one matching result. Similarly, a query for { a: { $gt: 1}, b: "tv" } must scan 6 documents, also to return one result. Consider the same index on a collection where a has nine values evenly distributed across the collection:
{ { { { { { { { { _id: _id: _id: _id: _id: _id: _id: _id: _id: ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), ObjectId(), a: a: a: a: a: a: a: a: a: 1, 2, 3, 4, 5, 6, 7, 8, 9, b: b: b: b: b: b: b: b: b: "ab" "cd" "ef" "jk" "lm" "no" "pq" "rs" "tv" } } } } } } } } }
If you query for { a: 2, b: "cd" }, MongoDB must scan only one document to fulll the query. The index and query are more selective because the values of a are evenly distributed and the query can select a specic document
33.1. Indexing Strategies
345
using the index. However, although the index on a is more selective, a query such as { a: still need to scan 4 documents. { $gt: 5 }, b: "tv" } would
If overall selectivity is low, and if MongoDB must read a number of documents to return results, then some queries may perform faster without indexes. To determine performance, see Measure Index Use (page 355). For a conceptual introduction to indexes in MongoDB see Indexing Overview (page 329).
346
CHAPTER 34
Index Tutorials
34.1 Indexing Operations

Indexes allow MongoDB to process and fulll queries quickly by creating small and efcient representations of the documents in a collection. The documents in this section outline specic tasks related to building and maintaining indexes for data in MongoDB collections. For a conceptual overview of MongoDB indexing, see the Indexing Overview (page 329) document. For strategies and practical approaches, see the Indexing Strategies (page 341) document.
34.1.1 Create an Index

Indexes allow MongoDB to process and fulll queries quickly by creating small and efcient representations of the documents in a collection. MongoDB creates an index on the _id eld of every collection by default, but allows users to create indexes for any collection using on any eld in a document. This tutorial describes how to create an index on a single eld. MongoDB also supports compound indexes (page 331), which are indexes on multiple elds. See Create a Compound Index (page 348) for instructions on building compound indexes. Build a Foreground Index on a Single Field To create an index, use db.collection.ensureIndex() (page 921) or a similar method from your driver. For example the following creates an index on the phone-number eld of the people collection:
db.people.ensureIndex( { "phone-number": 1 } )
ensureIndex() (page 921) only creates an index if an index of the same specication does not already exist. All indexes support and optimize the performance for queries that select on this eld. For queries that cannot use an index, MongoDB must scan all documents in a collection for documents that match the query. Examples If you create an index on the user_id eld in the records, this index is, the index will support the following query:
db.records.find( { user_id: 2 } )
However, the following query, on the profile_url eld is not supported by this index:
347
db.records.find( { profile_url: 2 } )
Additional Considerations If your collection holds a large amount of data, and your application needs to be able to access the data while building the index, consider building the index in the background, as described in Background Construction (page 336). To build indexes on replica sets, see the Build Indexes on Replica Sets (page 351) section for more information. Note: To build or rebuild indexes for a replica set see Build Indexes on Replica Sets (page 351). Some drivers may specify indexes, using NumberLong(1) rather than 1 as the specication. This does not have any affect on the resulting index. See also: Create a Compound Index (page 348), Indexing Operations (page 347) and Indexing Overview (page 329) for more information.
34.1.2 Create a Compound Index

Indexes allow MongoDB to process and fulll queries quickly by creating small and efcient representations of the documents in a collection. MongoDB supports indexes that include content on a single eld, as well as compound indexes (page 331) that include content from multiple elds. Continue reading for instructions and examples of building a compound index. Build a Compound Index To create a compound index (page 331) use an operation that resembles the following prototype:
db.collection.ensureIndex( { a: 1, b: 1, c: 1 } )
Example The following operation will create an index on the item, category, and price elds of the products collection:
db.products.ensureIndex( { item: 1, category: 1, price: 1 } )
Additional Considerations If your collection holds a large amount of data, and your application needs to be able to access the data while building the index, consider building the index in the background, as described in Background Construction (page 336). To build indexes on replica sets, see the Build Indexes on Replica Sets (page 351) section for more information. Note: To build or rebuild indexes for a replica set see Build Indexes on Replica Sets (page 351). Some drivers may specify indexes, using NumberLong(1) rather than 1 as the specication. This does not have any affect on the resulting index. See also: Create an Index (page 347), Indexing Operations (page 347) and Indexing Overview (page 329) for more information. 348 Chapter 34. Index Tutorials
34.1.3 Create a Unique Index

MongoDB allows you to specify a unique constraint (page 334) on an index. These constraints prevent applications from inserting documents that have duplicate values for the inserted elds. Additionally, if you want to create an index on a collection that has existing data that might have duplicate values for the indexed eld, you may chose combine unique enforcement with duplicate dropping (page 337). Unique Indexes To create a unique indexes (page 334), consider the following prototype:
db.collection.ensureIndex( { a: 1 }, { unique: true } )
For example, you may want to create a unique index on the "tax-id": of the accounts collection to prevent storing multiple account records for the same legal entity:
db.accounts.ensureIndex( { "tax-id": 1 }, { unique: true } )
The _id index (page 330) is a unique index. In some situations you may consider using _id eld itself for this kind of data rather than using a unique index on another eld. In many situations you will want to combine the unique constraint with the sparse option. When MongoDB indexes a eld, if a document does not have a value for a eld, the index entry for that item will be null. Since unique indexes cannot have duplicate values for a eld, without the sparse option, MongoDB will reject the second document and all subsequent documents without the indexed eld. Consider the following prototype.
db.collection.ensureIndex( { a: 1 }, { unique: true, sparse: true } )
You can also enforce a unique constraint on compound indexes (page 331), as in the following prototype:
db.collection.ensureIndex( { a: 1, b: 1 }, { unique: true } )
These indexes enforce uniqueness for the combination of index keys and not for either key individually. Drop Duplicates To force the creation of a unique index (page 334) index on a collection with duplicate values in the eld you are indexing you can use the dropDups option. This will force MongoDB to create a unique index by deleting documents with duplicate values when building the index. Consider the following prototype invocation of db.collection.ensureIndex() (page 921):
db.collection.ensureIndex( { a: 1 }, { dropDups: true } )
See the full documentation of duplicate dropping (page 337) for more information. Warning: Specifying { dropDups: tion. true } may delete data from your database. Use with extreme cau-
Refer to the ensureIndex() (page 921) documentation for additional index creation options.
34.1.4 Create a Sparse Index

Sparse indexes are like non-sparse indexes, except that they omit references to documents that do not include the indexed eld. For elds that are only present in some documents sparse indexes may provide a signicant space savings. See Sparse Indexes (page 334) for more information about sparse indexes and their use.
34.1. Indexing Operations
349
See also: Indexing Overview (page 329) and Indexing Operations (page 347) for more information. Prototype To create a sparse index (page 334) on a eld, use an operation that resembles the following prototype:
db.collection.ensureIndex( { a: 1 }, { sparse: true } )
Example The following operation, creates a sparse index on the users collection that only includes a document in the index if the twitter_name eld exists in a document.
db.users.ensureIndex( { twitter_name: 1 }, { sparse: true } )
The index excludes all documents that do not include the twitter_name eld. Considerations Note: Sparse indexes can affect the results returned by the query, particularly with respect to sorts on elds not included in the index. See the sparse index (page 334) section for more information.
34.1.5 Create a Hashed Index

New in version 2.4. Hashed indexes (page 335) compute a hash of the value of a eld in a collection and index the hashed value. These indexes permit equality queries and may be suitable shard keys for some collections. See also: Hashed Sharding (page 486) for more information about hashed indexes in sharded clusters, as well as Indexing Overview (page 329) and Indexing Operations (page 347) for more information about indexes. Operation To create a hashed index (page 335), specify hashed as the value of the index key, as in the following prototype: Example
db.collection.ensureIndex( { a: "hashed" } )
Considerations MongoDB supports hashed indexes of any single eld. The hashing function collapses sub-documents and computes the hash for the entire value, but does not support multi-key (i.e. arrays) indexes. You may not create compound indexes that have hashed index elds.
350
Chapter 34. Index Tutorials
34.1.6 Build Indexes on Replica Sets

Background index creation operations (page 336) become foreground indexing operations on secondary members of replica sets. The foreground index building process blocks all replication and read operations on the secondaries while they build the index. Secondaries will begin building indexes after the primary nishes building the index. In sharded clusters, the mongos (page 1032) will send ensureIndex() (page 921) to the primary members of the replica set for each shard, which then replicate to the secondaries after the primary nishes building the index. To minimize the impact of building an index on your replica set, use the following procedure to build indexes on secondaries: See also: Indexing Operations (page 347) and Indexing Overview (page 329) for more information. Considerations Warning: Ensure that your oplog is large enough to permit the indexing or re-indexing operation to complete without falling too far behind to catch up. See the oplog sizing (page 392) documentation for additional information. Note: This procedure does take one member out of the replica set at a time. However, this procedure will only affect one member of the set at a time rather than all secondaries at the same time.
Procedure Note: If you need to build an index in a sharded cluster, repeat the following procedure for each replica set that provides each shard.
Stop One Secondary
Stop the mongod (page 1021) process on one secondary. Restart the mongod (page 1021) process without the --replSet (page 1028) option and running on a different port. 1 This instance is now in standalone mode. For example, if your mongod (page 1021) normally runs with on the default port of 27017 with the --replSet (page 1028) option you would use the following invocation:
mongod --port 47017
Build the Index
Create the new index using the ensureIndex() (page 921) in the mongo (page 1036) shell, or comparable method in your driver. This operation will create or rebuild the index on this mongod (page 1021) instance For example, to create an ascending index on the username eld of the records collection, use the following mongo (page 1036) shell operation:
By running the mongod (page 1021) on a different port, you ensure that the other members of the replica set and all clients will not contact the member while you are building the index.
1
351
db.records.ensureIndex( { username: 1 } )
See also: Create an Index (page 347) and Create a Compound Index (page 348) for more information.
Restart the Program mongod
When the index build completes, start the mongod (page 1021) instance with the --replSet (page 1028) option on its usual port:
mongod --port 27017 --replSet rs0
Modify the port number (e.g. 27017) or the replica set name (e.g. rs0) as needed. Allow replication to catch up on this member.
Build Indexes on all Secondaries
For each secondary in the set, build an index according to the following steps: 1. Stop One Secondary (page 351) 2. Build the Index (page 351) 3. Restart the Program mongod (page 352)
Build the Index on the Primary
Finally, to build the index on the primary, begin by stepping down the primary. Use the rs.stepDown() (page 991) method in the mongo (page 1036) shell to cause the current primary to become a secondary graceful and allow the set to elect another member as primary. Then repeat the index building procedure, listed below, to build the index on the primary: 1. Stop One Secondary (page 351) 2. Build the Index (page 351) 3. Restart the Program mongod (page 352)
34.1.7 Build Indexes in the Background

By default, MongoDB builds indexes in the foreground, which means that these indexes block all other read and write operations to the database while the index builds. Background index construction (page 336) allows read and write operations to continue while building the index; however, these index builds take longer to complete and result in a larger index. See also: Indexing Overview (page 329) and Indexing Operations (page 347) for more information.
352
Procedure To create an index in the background, add the background argument to the ensureIndex() (page 921) operation, as in the following index:
db.collection.ensureIndex( { a: 1 }, { background: true } )
Consider the section on background index construction (page 336) for more information about these indexes and their implications.
34.1.8 Remove Indexes

To remove an index from a collection use the dropIndex() method and the following procedure. If you simply need to rebuild indexes you can use the process described in the Rebuild Indexes (page 353) document. See also: Indexing Operations (page 347) and Indexing Overview (page 329) for more information about indexes and indexing operations in MongoDB. Operations To remove an index, use the db.collection.dropIndex() method, as in the following example:
db.accounts.dropIndex( { "tax-id": 1 } )
This will remove the index on the "tax-id" eld in the accounts collection. The shell provides the following document after completing the operation:
{ "nIndexesWas" : 3, "ok" : 1 }
Where the value of nIndexesWas reects the number of indexes before removing this index. You can also use the db.collection.dropIndexes() (page 921) to remove all indexes, except for the _id index (page 330) from a collection. These shell helpers provide wrappers around the dropIndexes (page 864) database command. Your client library (page 555) may have a different or additional interface for these operations.
34.1.9 Rebuild Indexes

If you need to rebuild indexes for a collection you can use the db.collection.reIndex() (page 944) method to rebuild all indexes on a collection in a single operation. This operation drops all indexes, including the _id index (page 330), and then rebuilds all indexes. See also: Indexing Overview (page 329) and Indexing Operations (page 347). Process The operation takes the following form:
db.accounts.reIndex()
MongoDB will return the following document when the operation completes:
353
{ "nIndexesWas" : 2, "msg" : "indexes dropped for collection", "nIndexes" : 2, "indexes" : [ { "key" : { "_id" : 1, "tax-id" : 1 }, "ns" : "records.accounts", "name" : "_id_" } ], "ok" : 1 }
This shell helper provides a wrapper around the reIndex (page 867) database command. Your client library (page 555) may have a different or additional interface for this operation. Additional Considerations Note: To build or rebuild indexes for a replica set see Build Indexes on Replica Sets (page 351).
34.1.10 Monitor and Manage In Progress Index Creation and Building

To see the status of the indexing processes, you can use the db.currentOp() (page 971) method in the mongo (page 1036) shell. The value of the query eld and the msg eld will indicate if the operation is an index build. The msg eld also indicates the percent of the build that is complete. To terminate an ongoing index build, use the db.killOp() (page 981) method in the mongo (page 1036) shell. For more information see db.currentOp() (page 971). Changed in version 2.4: Before MongoDB 2.4, you could only terminate background index builds. After 2.4, you can terminate any index build, including foreground index builds.
34.1.11 Return a List of All Indexes

When performing maintenance you may want to check which indexes exist on a collection. Every index on a collection has a corresponding document in the system.indexes (page 1099) collection, and you can use standard queries (i.e. find() (page 924)) to list the indexes, or in the mongo (page 1036) shell, the getIndexes() (page 930) method to return a list of the indexes on a collection, as in the following examples. See also: Indexing Overview (page 329) and Indexing Operations (page 347) for more information about indexes in MongoDB and common index management operations. List all Indexes on a Collection To return a list of all indexes on a collection, use the db.collection.getIndexes() (page 930) method or a similar method for your driver.
354
For example, to view all indexes on the people collection:

db.people.getIndexes()
List all Indexes for a Database To return a list of all indexes on all collections in a database, use the following operation in the mongo (page 1036) shell:
db.system.indexes.find()
See system.indexes (page 1099) for more information about these documents.
34.1.12 Measure Index Use

Synopsis Query performance is a good general indicator of index use; however, for more precise insight into index use, MongoDB provides a number of tools that allow you to study query operations and observe index use for your database. See also: Indexing Overview (page 329), Indexing Strategies (page 341), and Indexing Operations (page 347) for more information. Operations
Return Query Plan with explain()
Append the explain() (page 953) method to any cursor (e.g. query) to return a document with statistics about the query process, including the index used, the number of documents scanned, and the time the query takes to process in milliseconds.
Control Index Use with hint()
Append the hint() (page 958) to any cursor (e.g. query) with the index as the argument to force MongoDB to use a specic index to fulll the query. Consider the following example:
db.people.find( { name: "John Doe", zipcode: { $gt: 63000 } } } ).hint( { zipcode: 1 } )
You can use hint() (page 958) and explain() (page 953) in conjunction with each other to compare the effectiveness of a specic index. Specify the $natural operator to the hint() (page 958) method to prevent MongoDB from using any index:
db.people.find( { name: "John Doe", zipcode: { $gt: 63000 } } } ).hint( { $natural: 1 } )
Instance Index Use Reporting
MongoDB provides a number of metrics of index use and operation that you may want to consider when analyzing index use for your database: In the output of serverStatus (page 889):
355
indexCounters (page 895) scanned (page 904) scanAndOrder (page 904) In the output of collStats (page 873): totalIndexSize (page 874) indexSizes (page 874) In the output of dbStats (page 877): dbStats.indexes (page 878) dbStats.indexSize (page 878)
34.1.13 Build Old Style Indexes

Important: Use this procedure only if you must have indexes that are compatible with a version of MongoDB earlier than 2.0. MongoDB version 2.0 introduced the {v:1} index format. MongoDB versions 2.0 and later support both the {v:1} format and the earlier {v:0} format. MongoDB versions prior to 2.0, however, support only the {v:0} format. If you need to roll back MongoDB to a version prior to 2.0, you must drop and re-create your indexes. To build pre-2.0 indexes, use the dropIndexes() (page 921) and ensureIndex() (page 921) methods. You cannot simply reindex the collection. When you reindex on versions that only support {v:0} indexes, the v elds in the index denition still hold values of 1, even though the indexes would now use the {v:0} format. If you were to upgrade again to version 2.0 or later, these indexes would not work. Example Suppose you rolled back from MongoDB 2.0 to MongoDB 1.8, and suppose you had the following index on the items collection:
{ "v" : 1, "key" : { "name" : 1 }, "ns" : "mydb.items", "name" : "name_1" }
The v eld tells you the index is a {v:1} index, which is incompatible with version 1.8. To drop the index, issue the following command:
db.items.dropIndex( { name : 1 } )
To recreate the index as a {v:0} index, issue the following command:

db.foo.ensureIndex( { name : 1 } , { v : 0 } )
See also: Index Performance Enhancements (page 1168).
356
CHAPTER 35
Geospatial Indexing
See Geospatial Indexes and Queries (page 357) for an introduction to geospatial indexing.
35.1 Geospatial Indexes and Queries

MongoDB offers a number of indexes and query mechanisms to handle geospatial information. This section introduces MongoDBs geospatial features.
35.1.1 Surfaces
Before storing your location data and writing queries, you must decide the type of surface to use to perform calculations. The type you choose affects how you store data, what type of index to build, and the syntax of your queries. MongoDB offers two surface types: Spherical To calculate geometry over an Earth-like sphere, store your location data on a spherical surface and use 2dsphere (page 362) index. Store your location data as GeoJSON objects with this coordinate-axis order: longitude, latitude. The coordinate reference system for GeoJSON uses the WGS84 datum. Flat To calculate distances on a Euclidean plane, store your location data as legacy coordinate pairs and use a 2d (page 359) index.
35.1.2 Location Data

If you choose spherical surface calculations, you store location data as GeoJSON objects (preferred). Queries on GeoJSON objects always calculate on a sphere. The default coordinate reference system for GeoJSON uses the WGS84 datum. New in version 2.4: The storage and querying of GeoJSON objects is new in version 2.4. Prior to version 2.4, all geospatial data was stored as coordinate pairs. MongoDB supports the following GeoJSON objects:
357
Point LineString Polygon Legacy coordinate pairs MongoDB supports spherical surface calculations on legacy coordinate pairs by converting the data to the GeoJSON Point type. If you choose at surface calculations, you can store data only as legacy coordinate pairs.
35.1.3 Query Operations

MongoDBs geospatial query operators let you query for: Inclusion. MongoDB can query for locations contained entirely within a specied polygon. Inclusion queries use the $geoWithin (page 778) operator. Intersection. MongoDB can query for locations that intersect with a specied geometry. These queries apply only to data on a spherical surface. These queries use the $geoIntersects (page 777) operator. Proximity. MongoDB can query for the points nearest to another point. Proximity queries use the $near (page 780) operator. The $near (page 780) operator requires a 2d or 2dsphere index.
35.1.4 Geospatial Indexes

MongoDB provides the following geospatial index types to support the geospatial queries: 2dsphere (page 362), which supports: Calculations on a sphere Both GeoJSON objects and legacy coordinate pairs A compound index with scalar index elds (i.e. ascending or descending) as a prex or sufx of the 2dsphere index eld New in version 2.4: 2dsphere indexes are not available before version 2.4. 2d (page 359), which supports: Calculations using at geometry Legacy coordinate pairs (i.e., geospatial points on a at coordinate system) A compound index with only one additional eld, as a sufx of the 2d index eld
35.1.5 Geospatial Indexes and Sharding

You cannot use a geospatial index as the shard key index. You can create and maintain a geospatial index on a sharded collection if using different elds as the shard key. Queries using $near (page 780) are not supported for sharded collections. Use geoNear (page 822) instead. You also can query for geospatial data using $geoWithin (page 778).
358
Chapter 35. Geospatial Indexing
35.1.6 Additional Resources

The following pages provide complete documentation for geospatial indexes and queries: 2d Indexes Use a 2d index for data stored as points on a two-dimensional plane. The 2d index is intended for legacy coordinate pairs used in MongoDB 2.2 and earlier. Use a 2d index if: your database has legacy location data from MongoDB 2.2 or earlier, and you do not intend to store any location data as GeoJSON objects. Do not use a 2d index if your location data includes GeoJSON objects. To index on both legacy coordinate pairs and GeoJSON objects, use a 2dsphere (page 362) index. The 2d index supports calculations on a at, Euclidean plane. The 2d index also supports distance-only calculations on a sphere, but for geometric calculations on a sphere, store data as GeoJSON objects and use the 2dsphere index type. A 2d index can reference two elds. The rst must be the location eld. A 2d compound index constructs queries that select rst on the location eld and second on the additional eld. If the location criteria selects a large number of documents, the additional criteria only lters the result set. The additional criteria does not result in a more targeted query. MongoDB allows one 2d index per collection. MongoDB also allows one 2dsphere (page 362) index per collection. Important: You cannot use a 2d index as a shard key when sharding a collection. However, you can create and maintain a geospatial index on a sharded collection by using a different eld as the shard key.
Store Points on a 2D Plane
To store location data as legacy coordinate pairs, use either an array (preferred):
loc : [ <longitude> , <latitude> ]
Or an embedded document:
loc : { lng : <longitude> , lat : <latitude> }
Arrays are preferred as certain languages do not guarantee associative map ordering. Whether as an array or document, if you use longitude and latitude, store coordinates in this order: longitude, latitude.
Create a 2d Index
To build a geospatial 2d index, use the ensureIndex() (page 921) method and specify 2d. Use the following syntax:
db.<collection>.ensureIndex( { <location field> : "2d" , <additional field> : <value> } , { <index-specification options> } )
The 2d index uses the following optional index-specication options:
35.1. Geospatial Indexes and Queries
359
{ min : <lower bound> , max : <upper bound> , bits : <bit precision> }
Dene Location Range for a 2d Index By default, a 2d index assumes longitude and latitude and has boundaries of -180 inclusive and 180 non-inclusive (i.e. [ -180 , 180 ]). If documents contain coordinate data outside of the specied range, MongoDB returns an error. Important: The default boundaries allow applications to insert documents with invalid latitudes greater than 90 or less than -90. The behavior of geospatial queries with such invalid points is not dened. On 2d indexes you can change the location range. You can build a 2d geospatial index with a location range other than the default. Use the min and max options when creating the index. Use the following syntax:
db.collection.ensureIndex( { <location field> : "2d" } , { min : <lower bound> , max : <upper bound> } )
Dene Location Precision for a 2d Index By default, a 2d index on legacy coordinate pairs uses 26 bits of precision, which is roughly equivalent to 2 feet or 60 centimeters of precision using the default range of -180 to 180. Precision is measured by the size in bits of the geohash values used to store location data. You can congure geospatial indexes with up to 32 bits of precision. Index precision does not affect query accuracy. The actual grid coordinates are always used in the nal query processing. Advantages to lower precision are a lower processing overhead for insert operations and use of less space. An advantage to higher precision is that queries scan smaller portions of the index to return results. To congure a location precision other than the default, use the bits option when creating the index. Use following syntax:
db.<collection>.ensureIndex( {<location field> : "<index type>"} , { bits : <bit precision> } )
For information on the internals of geohash values, see Calculation of Geohash Values for 2d Indexes (page 369).
Query a 2d Index
The following sections describe queries supported by the 2d index. For an overview of recommended geospatial queries, see Geospatial Query Compatibility (page 781). Points within a Shape Dened on a Flat Surface To select all legacy coordinate pairs found within a given shape on a at surface, use the $geoWithin (page 778) operator along with a shape operator. Use the following syntax:
db.<collection>.find( { <location field> : { $geoWithin : { $box|$polygon|$center : <coordinates> } } } )
The following queries for documents within a rectangle dened by [ 0 , 0 ] at the bottom left corner and by [ 100 , 100 ] at the top right corner.
db.places.find( { loc : { $geoWithin : { $box : [ [ 0 , 0 ] ,
360
[ 100 , 100 ] ] } } } )
The following queries for documents that are within the circle centered on [ -74 , 40.74 ] and with a radius of 10:
db.places.find( { loc: { $geoWithin : { $center : [ [-74, 40.74 ] , 10 ] } } } )
For syntax and examples for each shape, see the following: $box $polygon $center (denes a circle) Points within a Circle Dened on a Sphere MongoDB supports rudimentary spherical queries on at 2d indexes for legacy reasons. In general, spherical calculations should use a 2dsphere index, as described in 2dsphere Indexes (page 362). To query for legacy coordinate pairs in a spherical cap on a sphere, use $geoWithin (page 778) with the $centerSphere operator. Specify an array that contains: The grid coordinates of the circles center point The circles radius measured in radians. To calculate radians, see Calculate Distances in a 2d Index Using Spherical Geometry (page 367). Use the following syntax:
db.<collection>.find( { <location field> : { $geoWithin : { $centerSphere : [ [ <x>, <y> ] , <radius> ] } } } )
The following example query returns all documents within a 10-mile radius of longitude 88 W and latitude 30 N. The example converts distance to radians by dividing distance by the approximate radius of the earth, 3959 miles:
db.<collection>.find( { loc : { $geoWithin : { $centerSphere : [ [ 88 , 30 ] , 10 / 3959 ] } } } )
Proximity to a Point on a Flat Surface Proximity queries return the 100 legacy coordinate pairs closest to the dened point and sort the results by distance. Use either the $near (page 780) operator or geoNear (page 822) command. Both require a 2d index. The $near (page 780) operator uses the following syntax:
db.<collection>.find( { <location field> : { $near : [ <x> , <y> ] } } )
For examples, see $near (page 780). The geoNear (page 822) command uses the following syntax:
361
db.runCommand( { geoNear: <collection>, near: [ <x> , <y> ] } )
The geoNear (page 822) command offers more options and returns more information than does the $near (page 780) operator. To run the command, see geoNear (page 822). Exact Matches on a Flat Surface You can use the db.collection.find() (page 924) method to query for an exact match on a location. These queries use the following syntax:
db.<collection>.find( { <location field>: [ <x> , <y> ] } )
This query will return any documents with the value of [ <x> , <y> ]. 2dsphere Indexes New in version 2.4. A 2dsphere index supports queries that calculate geometries on an earth-like sphere. The index supports data stored as both GeoJSON objects and as legacy coordinate pairs. The index supports legacy coordinate pairs by converting the data to the GeoJSON Point type. The 2dsphere index supports all MongoDB geospatial queries: queries for inclusion, intersection and proximity. A compound (page 331) 2dsphere index can reference multiple location and non-location elds within a collections documents. You can arrange the elds in any order. The default datum for an earth-like sphere in MongoDB 2.4 is WGS84. Coordinate-axis order is longitude, latitude. MongoDB allows one 2dsphere index per collection. MongoDB also allows one 2d (page 359) index per collection. Important: You cannot use a 2dsphere index as a shard key when sharding a collection. However, you can create and maintain a geospatial index on a sharded collection by using a different eld as the shard key.
Store GeoJSON Objects
New in version 2.4. MongoDB supports the following GeoJSON objects: Point LineString Polygon In order to index GeoJSON data, you must store the data in a location eld that you name. The location eld contains a subdocument with a type eld specifying the GeoJSON object type and a coordinates eld specifying the objects coordinates. Always store coordinates in this order: longitude, latitude. Use the following syntax:
{ <location field> : { type : "<GeoJSON type>" , coordinates : <coordinates> } }
The following example stores a GeoJSON Point:
362
{ loc : { type : "Point" , coordinates : [ 40, 5 ] } }
The following example stores a GeoJSON LineString:

{ loc : { type : "LineString" , coordinates : [ [ 40 , 5 ] , [ 41 , 6 ] ] } }
The following example stores a GeoJSON Polygon with an exterior ring and no interior rings (or holes):
{ loc : { type : "Polygon" , coordinates : [ [ [ 0 , 0 ] , [ 3 , 6 ] , [ 6 , 1 ] , [ 0 , 0 ] ] ] } }
Create a 2dsphere Index
To create a geospatial index for GeoJSON-formatted data, use the ensureIndex() (page 921) method and set the value of the location eld for your collection to 2dsphere. A 2dsphere index can be a compound index (page 331) and does not require the location eld to be the rst eld indexed. To create the index use the following syntax:
db.points.ensureIndex( { <location field> : "2dsphere" } )
The following are four example commands for creating a 2dsphere index:
db.points.ensureIndex( db.points.ensureIndex( db.points.ensureIndex( db.points.ensureIndex( { { { { loc : "2dsphere" loc : "2dsphere" rating : 1 , loc loc : "2dsphere" } , : , ) type : 1 } ) "2dsphere" } ) rating : 1 , category : -1 } )
The rst example creates a simple geospatial index on the location eld loc. The second example creates a compound index where the second eld contains non-location data. The third example creates an index where the location eld is not the primary eld: the location eld does not have to be the rst eld in a 2dsphere index. The fourth example creates a compound index with three elds. You can include as many elds as you like in a 2dsphere index.
Query a 2dsphere Index
The following sections describe queries supported by the 2dsphere index. For an overview of recommended geospatial queries, see Geospatial Query Compatibility (page 781). GeoJSON Objects Bounded by a Polygon The $geoWithin (page 778) operator queries for location data found within a GeoJSON polygon. Your location data must be stored in GeoJSON format. Use the following syntax:
db.<collection>.find( { <location field> : { $geoWithin : { $geometry : { type : "Polygon" , coordinates : [ <coordinates> ] } } } } )
The following example selects all points and shapes that exist entirely within a GeoJSON polygon:
363
db.places.find( { loc : { $geoWithin : { $geometry : { type : "Polygon" , coordinates : [ [ [ [ [ [ ] ] } } } } )
0 3 6 0
, , , ,
0 6 1 0
] , ] , ] , ]
Intersections of GeoJSON Objects New in version 2.4. The $geoIntersects (page 777) operator queries for locations that intersect a specied GeoJSON object. A location intersects the object if the intersection is non-empty. This includes documents that have a shared edge. The $geoIntersects (page 777) operator uses the following syntax:
db.<collection>.find( { <location field> : { $geoIntersects : { $geometry : { type : "<GeoJSON object type>" , coordinates : [ <coordinates> ] } } } } )
The following example uses $geoIntersects (page 777) to select all indexed points and shapes that intersect with the polygon dened by the coordinates array.
db.places.find( { loc : { $geoIntersects : { $geometry : { type : "Polygon" , coordinates: [ [ [ [ [ [ ] ] } } } } )
0 3 6 0
, , , ,
0 6 1 0
] , ] , ] , ]
Proximity to a GeoJSON Point Proximity queries return the 100 points closest to the dened point and sorts the results by distance. A proximity query on GeoJSON data requires a 2dsphere index. To query for proximity to a GeoJSON point, use either the $near (page 780) operator or geoNear (page 822) command. Distance is in meters. The $near (page 780) uses the following syntax:
db.<collection>.find( { <location field> : { $near : { $geometry : { type : "Point" , coordinates : [ <longitude> , <latitude> ] } , $maxDistance : <distance in meters> } } } )
For examples, see $near (page 780). 364 Chapter 35. Geospatial Indexing
The geoNear (page 822) command uses the following syntax:

db.runCommand( { geoNear: <collection>, near: [ <x> , <y> ] } )
The geoNear (page 822) command offers more options and returns more information than does the $near (page 780) operator. To run the command, see geoNear (page 822). Points within a Circle Dened on a Sphere To select all grid coordinates in a spherical cap on a sphere, use $geoWithin (page 778) with the $centerSphere operator. Specify an array that contains: The grid coordinates of the circles center point The circles radius measured in radians. To calculate radians, see Calculate Distances in a 2d Index Using Spherical Geometry (page 367). Use the following syntax:
db.<collection>.find( { <location field> : { $geoWithin : { $centerSphere : [ [ <x>, <y> ] , <radius> ] } } } )
The following example queries grid coordinates and returns all documents within a 10 mile radius of longitude 88 W and latitude 30 N. The example converts the distance, 10 miles, to radians by dividing by the approximate radius of the earth, 3959 miles:
db.places.find( { loc : { $geoWithin : { $centerSphere : [ [ 88 , 30 ] , 10 / 3959 ] } } } )
Haystack Indexes A haystack index is a special index that is optimized to return results over small areas. Haystack indexes improve performance on queries that use at geometry. For queries that use spherical geometry, a 2dsphere index is a better option than a haystack index. 2dsphere indexes allow eld reordering; haystack indexes require the rst eld to be the location eld. Also, haystack indexes are only usable via commands and so always return all results at once. Haystack indexes create buckets of documents from the same geographic area in order to improve performance for queries limited to that area. Each bucket in a haystack index contains all the documents within a specied proximity to a given longitude and latitude. This document describes how to: Create a Haystack Index (page 365) Query a Haystack Index (page 366)
Create a Haystack Index
To build a haystack index, use the bucketSize option when creating the index. A bucketSize of 5 creates an index that groups location values that are within 5 units of the specied longitude and latitude. The bucketSize also determines the granularity of the index. You can tune the parameter to the distribution of your data so that in general
365
you search only very small regions. The areas dened by buckets can overlap. A document can exist in multiple buckets. A haystack index can reference two elds: the location eld and a second eld. The second eld is used for exact matches. Haystack indexes return documents based on location and an exact match on a single additional criterion. These indexes are not necessarily suited to returning the closest documents to a particular location. To build a haystack index, use the following syntax:
db.coll.ensureIndex( { <location field> : "geoHaystack" , <additional field> : 1 } , { bucketSize : <bucket value> } )
Example If you have a collection with documents that contain elds similar to the following:
{ _id : 100, pos: { lng : 126.9, lat : 35.2 } , type : "restaurant"} { _id : 200, pos: { lng : 127.5, lat : 36.1 } , type : "restaurant"} { _id : 300, pos: { lng : 128.0, lat : 36.7 } , type : "national park"}
The following operations create a haystack index with buckets that store keys within 1 unit of longitude or latitude.
db.places.ensureIndex( { pos : "geoHaystack", type : 1 } , { bucketSize : 1 } )
This index stores the document with an _id eld that has the value 200 in two different buckets: In a bucket that includes the document where the _id eld has a value of 100 In a bucket that includes the document where the _id eld has a value of 300 To query using a haystack index you use the geoSearch (page 823) command. See Query a Haystack Index (page 366). By default, queries that use a haystack index return 50 documents. Query a Haystack Index A haystack index is a special 2d geospatial index that is optimized to return results over small areas. To create a haystack index see Create a Haystack Index (page 365). To query a haystack index, use the geoSearch (page 823) command. You must specify both the coordinates and the additional eld to geoSearch (page 823). For example, to return all documents with the value restaurant in the type eld near the example point, the command would resemble:
db.runCommand( { geoSearch : "places" , search : { type: "restaurant" } , near : [-74, 40.74] , maxDistance : 10 } )
Note: Haystack indexes are not suited to queries for the complete list of documents closest to a particular location. The closest documents could be more distant compared to the bucket size. Note: Spherical query operations (page 367) are not currently supported by haystack indexes. The find() (page 924) method and geoNear (page 822) command cannot access the haystack index.
366
Calculate Distances in a 2d Index Using Spherical Geometry Note: While basic queries using spherical distance are supported by the 2d index, consider moving to a 2dsphere index if your data is primarily longitude and latitude. The 2d index supports queries that calculate distances on a Euclidean plane (at surface). The index also supports the following query operators and command that calculate distances using spherical geometry: $nearSphere (page 779) $centerSphere $near (page 780) geoNear (page 822) command with the { spherical: true } option.
Important: These three queries use radians for distance. Other query types do not. For spherical query operators to function properly, you must convert distances to radians, and convert from radians to the distances units used by your application. To convert: distance to radians: divide the distance by the radius of the sphere (e.g. the Earth) in the same units as the distance measurement. radians to distance: multiply the radian measure by the radius of the sphere (e.g. the Earth) in the units system that you want to convert the distance to. The radius of the Earth is approximately 3,959 miles or 6,371 kilometers. The following query would return documents from the places collection within the circle described by the center [ -74, 40.74 ] with a radius of 100 miles:
db.places.find( { loc: { $geoWithin: { $centerSphere: [ [ -74, 40.74 ] , 100 / 3959 ] } } } )
You may also use the distanceMultiplier option to the geoNear (page 822) to convert radians in the mongod (page 1021) process, rather than in your application code. See distance multiplier (page 368). The following spherical query, returns all documents in the collection places within 100 miles from the point [ -74, 40.74 ].
db.runCommand( { geoNear: "places", near: [ -74, 40.74 ], spherical: true } )
The output of the above command would be:

{ // [ ... ] "results" : [ { "dis" : 0.01853688938212826, "obj" : { "_id" : ObjectId( ... ) "loc" : [ -73, 40 ]
367
} } ], "stats" : { // [ ... ] "avgDistance" : 0.01853688938212826, "maxDistance" : 0.01853714811400047 }, "ok" : 1 }
Warning: Spherical queries that wrap around the poles or at the transition from -180 to 180 longitude raise an error. Note: While the default Earth-like bounds for geospatial indexes are between -180 inclusive, and 180, valid values for latitude are between -90 and 90.
Distance Multiplier
The distanceMultiplier option of the geoNear (page 822) command returns distances only after multiplying the results by an assigned value. This allows MongoDB to return converted values, and removes the requirement to convert units in application logic. Using distanceMultiplier in spherical queries provides results from the geoNear (page 822) command that do not need radian-to-distance conversion. The following example uses distanceMultiplier in the geoNear (page 822) command with a spherical (page 367) example:
db.runCommand( { geoNear: "places", near: [ -74, 40.74 ], spherical: true, distanceMultiplier: 3959 } )
The output of the above operation would resemble the following:

{ // [ ... ] "results" : [ { "dis" : 73.46525170413567, "obj" : { "_id" : ObjectId( ... ) "loc" : [ -73, 40 ] } } ], "stats" : { // [ ... ] "avgDistance" : 0.01853688938212826, "maxDistance" : 0.01853714811400047 }, "ok" : 1 }
368
2d Index Internals This document provides a more in-depth explanation of the internals of MongoDBs 2d geospatial indexes. This material is not necessary for normal operations or application development but may be useful for troubleshooting and for further understanding.
Calculation of Geohash Values for 2d Indexes
When you create a geospatial index on legacy coordinate pairs, MongoDB computes geohash values for the coordinate pairs within the specied location range (page 360) and then indexes the geohash values. To calculate a geohash value, recursively divide a two-dimensional map into quadrants. Then assign each quadrant a two-bit value. For example, a two-bit representation of four quadrants would be:
01 00 11 10
These two-bit values (00, 01, 10, and 11) represent each of the quadrants and all points within each quadrant. For a geohash with two bits of resolution, all points in the bottom left quadrant would have a geohash of 00. The top left quadrant would have the geohash of 01. The bottom right and top right would have a geohash of 10 and 11, respectively. To provide additional precision, continue dividing each quadrant into sub-quadrants. Each sub-quadrant would have the geohash value of the containing quadrant concatenated with the value of the sub-quadrant. The geohash for the upper-right quadrant is 11, and the geohash for the sub-quadrants would be (clockwise from the top left): 1101, 1111, 1110, and 1100, respectively. To calculate a more precise geohash, continue dividing the sub-quadrant and concatenate the two-bit identier for each division. The more bits in the hash identier for a given point, the smaller possible area that the hash can describe and the higher the resolution of the geospatial index.
Multi-location Documents for 2d Indexes
New in version 2.0: Support for multiple locations in a document. While 2d geospatial indexes do not support more than one set of coordinates in a document, you can use a multi-key index (page 332) to index multiple coordinate pairs in a single document. In the simplest example you may have a eld (e.g. locs) that holds an array of coordinates, as in the following example:
{ _id : ObjectId(...), locs : [ [ 55.5 , 42.3 ] , [ -74 , 44.74 ] , { lng : 55.5 , lat : 42.3 } ] }
The values of the array may be either arrays, as in [ 55.5, 42.3 ], or embedded documents, as in { lng : 55.5 , lat : 42.3 }. You could then create a geospatial index on the locs eld, as in the following:
db.places.ensureIndex( { "locs": "2d" } )
369
You may also model the location data as a eld inside of a sub-document. In this case, the document would contain a eld (e.g. addresses) that holds an array of documents where each document has a eld (e.g. loc:) that holds location coordinates. For example:
{ _id : ObjectId(...), name : "...", addresses : [ { context loc : [ } , { context loc : [ } ] }
: "home" , 55.5, 42.3 ]
: "home", -74 , 44.74 ]
You could then create the geospatial index on the addresses.loc eld as in the following example:
db.records.ensureIndex( { "addresses.loc": "2d" } )
For documents with multiple coordinate values, queries may return the same document multiple times if more than one indexed coordinate pair satises the query constraints. Use the uniqueDocs parameter to geoNear (page 822) or the $uniqueDocs operator with $geoWithin (page 778). To include the location eld with the distance eld in multi-location document queries, specify includeLocs: true in the geoNear (page 822) command. See also: Geospatial Query Compatibility (page 781)
370
CHAPTER 36
Text Indexing
New in version 2.4: text indexes were added in 2.4 as a beta feature.
36.1 Text Search

New in version 2.4.
36.1.1 Overview
Text search supports the search of string content in documents of a collection. Text search introduces a new text (page 338) index type and a new text (page 832) command. The text search process: tokenizes and stems the search term(s) during both the index creation and the text command execution. assigns a score to each document that contains the search term in the indexed elds. The score determines the relevance of a document to a given search query. By default, text (page 832) command returns at most the top 100 matching documents as determined by the scores. Important: Before you can create a text index or run the text command (page 378), you need to manually enable the text search. See Enable Text Search (page 371) for information on how to enable the text search feature.
Enable Text Search New in version 2.4. The text search (page 371) is currently a beta feature. As a beta feature: You need to explicitly enable the feature before creating a text index (page 372) or using the text (page 832) command. To enable text search on replica sets (page 387) and sharded clusters (page 485), you need to enable on each and every mongod (page 1021) for replica sets and on each and every mongos (page 1032) for sharded clusters. Warning: Do not enable or use text search on production systems. Text indexes have signicant storage requirements and performance costs. See text Indexes (page 338) for more information.
371
You can enable the text search feature at startup with the textSearchEnabled (page 1095) parameter:
mongod --setParameter textSearchEnabled=true
You may prefer to set the textSearchEnabled (page 1095) parameter in the conguration le (page 1078). Additionally, you can enable the feature in the mongo (page 1036) shell with the setParameter (page 869) command. This command does not propagate from the primary to the secondaries. You must enable on each and every mongod (page 1021) for replica sets. Note: You must set the parameter every time you start the server. You may prefer to add the parameter to the conguration les (page 1078).
36.1.2 Storage Requirements and Performance Costs

text indexes have the following storage requirements and performance costs: text indexes change the space allocation method for all future record allocations in a collection to usePowerOf2Sizes (page 857). text indexes can be large. They contain one index entry for each unique post-stemmed word in each indexed eld for each document inserted. Building a text index is very similar to building a large multi-key index and will take longer than building a simple ordered (scalar) index on the same data. When building a large text index on an existing collection, ensure that you have a sufciently high limit on open le descriptors. See the recommended settings (page 109). text indexes will impact insertion throughput because MongoDB must add an index entry for each unique post-stemmed word in each indexed eld of each new source document. Additionally, text indexes do not store phrases or information about the proximity of words in the documents. As a result, phrase queries will run much more effectively when the entire collection ts in RAM.
36.1.3 Create a text Index

To perform text search, create a text index on the eld or elds whose value is a string or an array of string elements. To create a text index, use the db.collection.ensureIndex() (page 921) method with a document that contains eld and value pairs where the value is the string literal text. Important: Text indexes have signicant storage requirements and performance costs. See Storage Requirements and Performance Costs (page 372) for more information. A collection can have at most one text index. The following tutorials offer examples on text index creation patterns: Create text Index on Multiple Fields You can create a text index on the eld or elds whose value is a string or an array of string elements. When creating a text index on multiple elds, you can specify the individual elds or you can wildcard specier ($**).
372
Chapter 36. Text Indexing
Index Specic Fields
The following example creates a text index on the elds subject and content:
db.collection.ensureIndex( { subject: "text", content: "text" } )
This text index catalogs all string data in the subject eld and the content eld, where the eld value is either a string or an array of string elements.
Index All Fields
To allow for text search on all elds with string content, use the wildcard specier ($**) to index all elds that contain string content. The following example indexes any string value in the data of every eld of every document in collection and names the index TextIndex:
db.collection.ensureIndex( { "$**": "text" }, { name: "TextIndex" } )
Specify a Language for Text Index The default language associated with the indexed data determines the list of stop words and the rules for the stemmer and tokenizer. The default language for the indexed data is english. To specify a different language, use the default_language option when creating the text index. See Text Search Languages (page 836) for the languages available for default_language. The following example creates a text index on the content eld and sets the default_language to spanish:
db.collection.ensureIndex( { content : "text" }, { default_language: "spanish" } )
See also: Create a text Index on a Multi-language Collection (page 374) Specify text Index Name to Avoid Name Length Limit The default name for the index consists of each indexed eld name concatenated with _text. For example, the following command creates a text index on the elds content, users.comments, and users.profiles:
db.collection.ensureIndex( { content: "text", "users.comments": "text", "users.profiles": "text"
36.1. Text Search
373
} )
The default name for the index is:

"content_text_users.comments_text_users.profiles_text"
To avoid creating an index with a name that exceeds the index name length limit (page 1106), you can pass the name option to the db.collection.ensureIndex() (page 921) method:
db.collection.ensureIndex( { content: "text", "users.comments": "text", "users.profiles": "text" }, { name: "MyTextIndex" } )
Note: To drop the text index, use the index name. db.collection.getIndexes() (page 930).
To get the name of an index, use
Create a text Index on a Multi-language Collection

Specify the Index Language within the Document
If a collection contains documents that are in different languages, include a eld in the documents that contain the language to use: If you include a eld named language in the document, by default, the ensureIndex() (page 921) method will use the value of this eld to override the default language. To use a eld with a name other than language, you must specify the name of this eld to the ensureIndex() (page 921) method with the language_override option. See Text Search Languages (page 836) for a list of supported languages.
Include the language Field
Include a eld language that species the language to use for the individual documents. For example, the documents of a multi-language collection quotes contain the eld language:
{ _id: 1, language: "portuguese", quote: "A sorte protege os audazes" } { _id: 2, language: "spanish", quote: "Nada hay ms surreal que la realidad." } { _id: 3, language: "english", quote: "is this a dagger which I see before me" }
Create a text index on the eld quote:

db.quotes.ensureIndex( { quote: "text" } )
For the documents that contain the language eld, the text index uses that language to determine the stop words and the rules for the stemmer and the tokenizer.
374
For documents that do not contain the language eld, the index uses the default language, which is English, to determine the stop words and rules for the stemmer and the tokenizer. For example, the Spanish word que is a stop word. So the following text (page 832) command would not match any document:
db.quotes.runCommand( "text", { search: "que", language: "spanish" } )
Use any Field to Specify the Language for a Document
Include a eld that species the language to use for the individual documents. To use a eld with a name other than language, include the language_override option when creating the index. For example, the documents of a multi-language collection quotes contain the eld idioma:
{ _id: 1, idioma: "portuguese", quote: "A sorte protege os audazes" } { _id: 2, idioma: "spanish", quote: "Nada hay ms surreal que la realidad." } { _id: 3, idioma: "english", quote: "is this a dagger which I see before me" }
Create a text index on the eld quote with the language_override option:
db.quotes.ensureIndex( { quote : "text" }, { language_override: "idioma" } )
For the documents that contain the idioma eld, the text index uses that language to determine the stop words and the rules for the stemmer and the tokenizer. For documents that do not contain the idioma eld, the index uses the default language, which is English, to determine the stop words and rules for the stemmer and the tokenizer. For example, the Spanish word que is a stop word. So the following text (page 832) command would not match any document:
db.quotes.runCommand( "text", { search: "que", language: "spanish" } )
Control Results of Text Search with Weights By default, the text (page 832) command returns matching documents based on scores, from highest to lowest. For a text index, the weight of an indexed eld denotes the signicance of the eld relative to the other indexed elds in terms of the score. The score for a given word in a document is derived from the weighted sum of the frequency for each of the indexed elds in that document. The default weight is 1 for the indexed elds. To adjust the weights for the indexed elds, include the weights option in the db.collection.ensureIndex() (page 921) method. Warning: Choose the weights carefully in order to prevent the need to reindex. A collection blog has the following documents:
{ _id: 1, content: "This morning I had a cup of coffee.", about: "beverage", keywords: [ "coffee" ] } { _id: 2, content: "Who doesnt like cake?",
36.1. Text Search
375
about: "food", keywords: [ "cake", "food", "dessert" ] }
To create a text index with different eld weights for the content eld and the keywords eld, include the weights option to the ensureIndex() (page 921) method. For example, the following command creates an index on three elds and assigns weights to two of the elds:
db.blog.ensureIndex( { content: "text", keywords: "text", about: "text" }, { weights: { content: 10, keywords: 5, }, name: "TextIndex" } )
The text index has the following elds and weights: content has a weight of 10, keywords has a weight of 5, and about has the default weight of 1. These weights denote the relative signicance of the indexed elds to each other. For instance, a term match in the content eld has: 2 times (i.e. 10:5) the impact as a term match in the keywords eld and 10 times (i.e. 10:1) the impact as a term match in the about eld. Limit the Number of Index Entries Scanned for Text Search The text (page 832) command includes the filter option to further restrict the results of a text search. For a filter that species equality conditions, this tutorial demonstrates how to perform text searches on only those documents that match the filter conditions, as opposed to performing a text search rst on all the documents and then matching on the filter condition. Consider a collection inventory that contains the following documents:
{ { { { { { _id: _id: _id: _id: _id: _id: 1, 2, 3, 4, 5, 6, dept: dept: dept: dept: dept: dept: "tech", description: "a "tech", description: "a "kitchen", description: "kitchen", description: "food", description: "a "food", description: "a fun green computer" } wireless red mouse" } "a green placemat" } "a red peeler" } green apple" } red potato" }
A common use case is to perform text searches by individual departments, such as:
db.inventory.runCommand( "text", { search: "green", filter: { dept : "kitchen" }
376
} )
To limit the text search to scan only those documents within a specic dept, create a compound index that species an ascending/descending index key on the eld dept and a text index key on the eld description:
db.inventory.ensureIndex( { dept: 1, description: "text" } )
Important: The ascending/descending index keys must be listed before, or prex, the text index keys. By prexing the text index elds with ascending/descending index elds, MongoDB will only index documents that have the prex elds. You cannot include multi-key (page 332) index elds or geospatial (page 338) index elds. The text (page 832) command must include the filter option that species an equality condition for the prex elds. Then, the text search within a particular department will limit the scan of indexed documents. For example, the following text (page 832) command scans only those documents with dept equal to kitchen:
db.inventory.runCommand( "text", { search: "green", filter: { dept : "kitchen" } } )
The returned result includes the statistics that shows that the command scanned 1 document, as indicated by the nscanned eld:
{ "queryDebugString" : "green||||||", "language" : "english", "results" : [ { "score" : 0.75, "obj" : { "_id" : 3, "dept" : "kitchen", "description" : "a green placemat" } } ], "stats" : { "nscanned" : 1, "nscannedObjects" : 0, "n" : 1, "nfound" : 1, "timeMicros" : 211 },
36.1. Text Search
377
"ok" : 1 }
For more information on the result set, see Output (page 836). Return Text Queries Using Only a text Index To create a text index that can cover queries (page 342): 1. Append scalar index elds to a text index, as in the following example which species an ascending index key on username:
db.collection.ensureIndex( { comments: "text", username: 1 } )
Warning: You cannot include multi-key (page 332) index eld or geospatial (page 338) index eld. 2. Use the project option in the text (page 832) to return only the elds in the index, as in the following:
db.quotes.runCommand( "text", { search: "tomorrow", project: { username: 1, _id: 0 } } )
Note: By default, the _id eld is included in the result set. Since the example index did not include the _id eld, you must explicitly exclude the eld in the project document.
36.1.4 text Command

The text (page 832) command can search for words and phrases. The command matches on the complete stemmed words. For example, if a document eld contains the word blueberry, a search on the term blue will not match the document. However, a search on either blueberry or blueberries will match. For information and examples on various text search patterns, see Search String Content for Text (page 378). Search String Content for Text In 2.4, you can enable the text search feature to create text indexes and issue text queries using the text (page 832). The following tutorial offers various query patterns for using the text search feature. The examples in this tutorial use a collection quotes that has a text index on the elds quote that contains a string and related_quotes that contains an array of string elements.
Search for a Term
The following command searches for the word TOMORROW:

db.quotes.runCommand( "text", { search: "TOMORROW" } )
378
Because text (page 832) command is case-insensitive, the text search will match the following document in the quotes collection:
{ "_id" : ObjectId("50ecef5f8abea0fda30ceab3"), "quote" : "tomorrow, and tomorrow, and tomorrow, creeps in this petty pace", "related_quotes" : [ "is this a dagger which I see before me", "the handle toward my hand?" ], "src" : { "title" : "Macbeth", "from" : "Act V, Scene V" }, "speaker" : "macbeth" }
Match Any of the Search Terms
If the search string is a space-delimited text, text (page 832) command performs a logical OR search on each term and returns documents that contains any of the terms. For example, the search string "tomorrow largo" searches for the term tomorrow OR the term largo:
db.quotes.runCommand( "text", { search: "tomorrow largo" } )
The command will match the following documents in the quotes collection:
{ "_id" : ObjectId("50ecef5f8abea0fda30ceab3"), "quote" : "tomorrow, and tomorrow, and tomorrow, creeps in this petty pace", "related_quotes" : [ "is this a dagger which I see before me", "the handle toward my hand?" ], "src" : { "title" : "Macbeth", "from" : "Act V, Scene V" }, "speaker" : "macbeth" } { "_id" : ObjectId("50ecf0cd8abea0fda30ceab4"), "quote" : "Es tan corto el amor y es tan largo el olvido.", "related_quotes" : [ "Como para acercarla mi mirada la busca.", "Mi corazn la busca, y ella no est conmigo." ], "speaker" : "Pablo Neruda", "src" : { "title" : "Veinte poemas de amor y una cancin desesperada", "from" : "Poema 20" } }
36.1. Text Search
379
Match Phrases
To match the exact phrase that includes a space(s) as a single term, escape the quotes. For example, the following command searches for the exact phrase "and tomorrow":
db.quotes.runCommand( "text", { search: "\"and tomorrow\"" } )
If the search string contains both phrase and individual terms, the text (page 832) command performs a compound logical AND of the phrase with the compound logical OR of the single terms For example, the following command contains a search string that contains the individual terms corto and largo as well as the phrase \"and tomorrow\":
db.quotes.runCommand( "text", { search: "corto largo \"and tomorrow\"" } )
The text (page 832) command performs the equivalent to the following logical operation:
(corto OR largo OR tomorrow) AND ("and tomorrow")
Match Some Words But Not Others
A negated term is a term that is prexed by a minus sign -. If you negate a term, the text (page 832) command will exclude the documents that contain those terms from the results. Note: If the search text contains only negated terms, the text (page 832) command will not return any results. The following example returns those documents that contain the term tomorrow but not the term petty.
db.quotes.runCommand( "text" , { search: "tomorrow -petty" } )
Limit the Number of Matching Documents in the Result Set
Note: The result from the text (page 832) command must t within the maximum BSON Document Size (page 1105). By default, the text (page 832) command will return up to 100 matching documents, from highest to lowest scores. To override this default limit, use the limit option in the text (page 832) command, as in the following example:
db.quotes.runCommand( "text", { search: "tomorrow", limit: 2 } )
The text (page 832) command will return at most 2 of the highest scoring results. The limit can be any number as long as the result set ts within the maximum BSON Document Size (page 1105).
Specify Which Fields to Return in the Result Set
In the text (page 832) command, use the project option to specify the elds to include (1) or exclude (0) in the matching documents. Note: The _id eld is always returned unless explicitly excluded in the project document. The following example returns only the _id eld and the src eld in the matching documents: 380 Chapter 36. Text Indexing
db.quotes.runCommand( "text", { search: "tomorrow", project: { "src": 1 } } )
Search with Additional Query Conditions
The text (page 832) command can also use the filter option to specify additional query conditions. The following example will return the documents that contain the term tomorrow AND the speaker is macbeth:
db.quotes.runCommand( "text", { search: "tomorrow", filter: { speaker : "macbeth" } } )
See also: Limit the Number of Index Entries Scanned for Text Search (page 376)
Search for Text in Specic Languages
You can specify the language that determines the tokenization, stemming, and removal of stop words, as in the following example:
db.quotes.runCommand( "text", { search: "amor", language: "spanish" } )
See also: Create a text Index on a Multi-language Collection (page 374) See Text Search Languages (page 836) for a list of supported languages.
Text Search Output
The text (page 832) command returns a document that contains the result set. See Output (page 836) for information on the output.
36.1.5 Text Search Output

The text (page 832) command returns a document that contains the result set. See Output (page 836) for information on the output.
36.1. Text Search
381
382
Part VIII
Replication
383
Database replication ensures redundancy, backup, and automatic failover. Replication occurs through groups of servers known as replica sets.
385
386
CHAPTER 37
Replica Set Use and Operation
Consider these higher level introductions to replica sets:
37.1 Replica Set Fundamental Concepts

A MongoDB replica set is a cluster of mongod (page 1021) instances that replicate amongst one another and ensure automated failover. Most replica sets consist of two or more mongod (page 1021) instances with at most one of these designated as the primary and the rest as secondary members. Clients direct all writes to the primary, while the secondary members replicate from the primary asynchronously. Database replication with MongoDB adds redundancy, helps to ensure high availability, simplies certain administrative tasks such as backups, and may increase read capacity. Most production deployments use replication. If youre familiar with other database systems, you may think about replica sets as a more sophisticated form of traditional master-slave replication. 1 In master-slave replication, a master node accepts writes while one or more slave nodes replicate those write operations and thus maintain data sets identical to the master. For MongoDB deployments, the member that accepts write operations is the primary, and the replicating members are secondaries. MongoDBs replica sets provide automated failover. If a primary fails, the remaining members will automatically try to elect a new primary. A replica set can have up to 12 members, but only 7 members can have votes. For information regarding non-voting members, see non-voting members (page 389) See also: The Replication (page 385) index for a list of the documents in this manual that describe the operation and use of replica sets.
37.1.1 Member Congurations

You can congure replica set members in a variety of ways, as listed here. In most cases, members of a replica set have the default proprieties.
1 MongoDB also provides conventional master/slave replication. Master/slave replication operates by way of the same mechanism as replica sets, but lacks the automatic failover capabilities. While replica sets are the recommended solution for production, a replica set can support only 12 members in total. If your deployment requires more than 11 slave members, youll need to use master/slave replication.
387
Secondary-Only Members These members have data but cannot become primary under any circumstance. To congure a member to be secondaryonly, see Prevent Replica Set Member from Becoming Primary (page 453). Hidden Members These members cannot become primary and are invisible to client applications. However, hidden members do vote in elections (page 389). Hidden members are ideal for instances that will have signicantly different usage patterns than the other members and require separation from normal trafc. Typically, hidden members provide reporting, dedicated backups, and dedicated read-only testing and integration support. To congure a member to be a hidden member, see Congure a Replica Set Member as Hidden (page 452). Delayed Members Delayed members copy and apply operations from the primarys oplog with a specied delay. If a member has a delay of one hour, then the latest entry in this members oplog will not be more recent than one hour old, and the state of data for the member will reect the state of the set an hour earlier. Example If the current time is 09:52 and the secondary is a delayed by an hour, no operation will be more recent than 08:52. Delayed members may help recover from various kinds of human error. Such errors may include inadvertently deleted databases or botched application upgrades. Consider the following factors when determining the amount of slave delay to apply: Ensure that the length of the delay is equal to or greater than your maintenance windows. The size of the oplog is sufcient to capture more than the number of operations that typically occur in that period of time. For more information on oplog size, see Oplog (page 392). Delayed members must have a priority set to 0 to prevent them from becoming primary in their replica sets. Also these members should be hidden (page 388) to prevent your application from seeing or querying this member. To congure a member to be a delayed member, see Congure a Delayed Replica Set Member (page 451). Arbiters These members have no data and exist solely to participate in elections (page 389). Arbiters have the following interactions with the rest of the replica set: Credential exchanges that authenticate the arbiter with the replica set. All MongoDB processes within a replica set use keyles. These exchanges are encrypted. MongoDB only transmits the authentication credentials in a cryptographically secure exchange, and encrypts no other exchange. Exchanges of replica set conguration data and of votes. These are not encrypted. If your MongoDB deployment uses SSL, then all communications between arbiters and the other members of the replica set are secure. See the documentation Connect to MongoDB with SSL (page 75) for more information. As with all MongoDB components, run arbiters on secure networks. To add an arbiter to the replica set, see Add an Arbiter to Replica Set (page 448). 388 Chapter 37. Replica Set Use and Operation
Non-Voting Members These members do not vote in elections. Non-voting members are only used for larger sets with more than 12 members. To congure a member as non-voting, see Congure a Non-Voting Replica Set Member (page 452).
37.1.2 Failover and Recovery

Replica sets feature automated failover. If the primary goes ofine or becomes unresponsive and a majority of the original set members can still connect to each other, the set will elect a new primary. While failover is automatic, replica set administrators should still understand exactly how this process works. This section below describe failover in detail. In most cases, failover occurs without administrator intervention seconds after the primary either steps down, becomes inaccessible, or becomes otherwise ineligible to act as primary. If your MongoDB deployment does not failover according to expectations, consider the following operational errors: No remaining member is able to form a majority. This can happen as a result of network partitions that render some members inaccessible. Design your deployment to ensure that a majority of set members can elect a primary in the same facility as core application systems. No member is eligible to become primary. Members must have a priority setting greater than 0, have a state that is less than ten seconds behind the last operation to the replica set, and generally be more up to date than the voting members. In many senses, rollbacks (page 390) represent a graceful recovery from an impossible failover and recovery situation. Rollbacks occur when a primary accepts writes that other members of the set do not successfully replicate before the primary steps down. When the former primary begins replicating again it performs a rollback. If the operations replicate to another member and that member remains available and accessible to a majority of the replica set, there will be no rollback. Rollbacks remove those operations from the instance that were never replicated to the set so that the data set is in a consistent state. The mongod (page 1021) program writes rolled back data to a BSON le that you can view using bsondump (page 1051), applied manually using mongorestore (page 1048). You can prevent rollbacks using a replica acknowledged (page 399) write concern. These write operations require not only the primary to acknowledge the write operation, sometimes even the majority of the set to conrm the write operation before returning. See also: The Elections (page 389) section in the Replica Set Fundamental Concepts (page 387) document, and the Election Internals (page 409) section in the Replica Set Internals and Behaviors (page 408) document. Elections When any failover occurs, an election takes place to decide which member should become primary. Elections provide a mechanism for the members of a replica set to autonomously select a new primary without administrator intervention. The election allows replica sets to recover from failover situations very quickly and robustly. Whenever the primary becomes unreachable, the secondary members trigger an election. The rst member to receive votes from a majority of the set will become primary. The most important feature of replica set elections is that a majority of the original number of members in the replica set must be present for election to succeed. If you have a three-member replica set, the set can elect a primary when two or three members can connect to each other. If two members in the replica go ofine, then the remaining member will remain a secondary.
37.1. Replica Set Fundamental Concepts
389
Note: When the current primary steps down and triggers an election, the mongod (page 1021) instances will close all client connections. This ensures that the clients maintain an accurate view of the replica set and helps prevent rollbacks. Members on either side of a network partition cannot see each other when determining whether a majority is available to hold an election. That means that if a primary steps down and neither side of the partition has a majority on its own, the set will not elect a new primary and the set will become read only. To avoid this situation, attempt to place a majority of instances in one data center with a minority of instances in a secondary facility. For more information on elections and failover, see the Failover and Recovery (page 389) section in the Troubleshoot Replica Sets (page 444) document. Member Priority In a replica set, every member has a priority, that helps determine eligibility for election (page 389) to primary. By default, all members have a priority of 1, unless you modify the priority (page 464) value. All members have a single vote in elections. Warning: Always congure the priority (page 464) value to control which members will become primary. Do not congure votes (page 465) except to permit more than 7 secondary members. For more information on member priorities, see the Adjust Priority for Replica Set Member (page 428) document.
37.1.3 Consistency
This section provides an overview of the concepts that underpin database consistency and the MongoDB mechanisms to ensure that users have access to consistent data. In MongoDB, all read operations issued to the primary of a replica set are consistent with the last write operation. If clients congure the read preference to permit secondary reads, read operations cannot return from secondary members that have not replicated more recent updates or operations. In these situations the query results may reect a previous state. This behavior is sometimes characterized as eventual consistency because the secondary members state will eventually reect the primarys state and MongoDB cannot guarantee strict consistency for read operations from secondary members. There is no way to guarantee consistency for reads from secondary members, except by conguring the client and driver to ensure that write operations succeed on all members before completing successfully. Rollbacks In some failover situations primaries will have accepted write operations that have not replicated to the secondaries after a failover occurs. This case is rare and typically occurs as a result of a network partition with replication lag. When this member (the former primary) rejoins the replica set and attempts to continue replication as a secondary the former primary must revert these operations or roll back these operations to maintain database consistency across the replica set. MongoDB writes the rollback data to a BSON le in the databases dbpath (page 1081) directory. Use bsondump (page 1051) to read the contents of these rollback les and then manually apply the changes to the new primary. There
390
Chapter 37. Replica Set Use and Operation
is no way for MongoDB to appropriately and fairly handle rollback situations automatically. Therefore you must intervene manually to apply rollback data. Even after the member completes the rollback and returns to secondary status, administrators will need to apply or decide to ignore the rollback data. MongoDB writes rollback data to a rollback/ folder within the dbpath (page 1081) directory to les with lenames in the following form:
<database>.<collection>.<timestamp>.bson
For example:
records.accounts.2011-05-09T18-10-04.0.bson
The best strategy for avoiding all rollbacks is to ensure write propagation (page 399) to all or some of the members in the set. Using these kinds of policies prevents situations that might create rollbacks. Warning: A mongod (page 1021) instance will not rollback more than 300 megabytes of data. If your system needs to rollback more than 300 MB, you will need to manually intervene to recover this data. If this is the case, you will nd the following line in your mongod (page 1021) log:
[replica set sync] replSet syncThread: 13410 replSet too much data to roll back
In these situations you will need to manually intervene to either save data or to force the member to perform an initial sync from a current member of the set by deleting the content of the existing dbpath (page 1081) directory. For more information on failover, see: The Failover and Recovery (page 389) section in this document. The Failover and Recovery (page 389) section in the Replica Set Administration (page 419) document. Application Concerns Client applications are indifferent to the conguration and operation of replica sets. While specic conguration depends to some extent on the client drivers (page 555), there is often minimal or no difference between applications using replica sets or standalone instances. There are two major concepts that are important to consider when working with replica sets: 1. Write Concern (page 398). Write concern sends a MongoDB client a response from the server to conrm successful write operations. In replica sets you can congure replica acknowledged (page 399) write concern to ensure that secondary members of the set have replicated operations before the write returns. 2. Read Preference (page 402) By default, read operations issued against a replica set return results from the primary. Users may congure read preference on a per-connection basis to prefer that read operations return on the secondary members. Read preference and write concern have particular consistency (page 390) implications. For a more detailed discussion of application concerns, see Replica Set Considerations and Behaviors for Applications and Development (page 398).
37.1.4 Administration and Operations

This section provides a brief overview of concerns relevant to administrators of replica set deployments. For more information on replica set administration, operations, and architecture, see: 37.1. Replica Set Fundamental Concepts 391
Deploy a Replica Set (page 419) Add Members to a Replica Set (page 424) Remove Members from Replica Set (page 427) Replace a Replica Set Member (page 427) Adjust Priority for Replica Set Member (page 428) Resync a Member of a Replica Set (page 428) Congure a Secondarys Sync Target (page 454) Congure a Delayed Replica Set Member (page 451) Congure a Replica Set Member as Hidden (page 452) Congure a Non-Voting Replica Set Member (page 452) Prevent Replica Set Member from Becoming Primary (page 453) Replica Set Architectures and Deployment Patterns (page 395) Oplog The oplog (operations log) is a special capped collection that keeps a rolling record of all operations that modify that data stored in your databases. MongoDB applies database operations on the primary and then records the operations on the primarys oplog. The secondary members then replicate this log and apply the operations to themselves in an asynchronous process. All replica set members contain a copy of the oplog, allowing them to maintain the current state of the database. Operations in the oplog are idempotent. By default, the size of the oplog is as follows: For 64-bit Linux, Solaris, FreeBSD, and Windows systems, MongoDB will allocate 5% of the available free disk space to the oplog. If this amount is smaller than a gigabyte, then MongoDB will allocate 1 gigabyte of space. For 64-bit OS X systems, MongoDB allocates 183 megabytes of space to the oplog. For 32-bit systems, MongoDB allocates about 48 megabytes of space to the oplog. Before oplog creation, you can specify the size of your oplog with the oplogSize (page 1088) option. After you start a replica set member for the rst time, you can only change the size of the oplog by using the Change the Size of the Oplog (page 435) tutorial. In most cases, the default oplog size is sufcient. For example, if an oplog that is 5% of free disk space lls up in 24 hours of operations, then secondaries can stop copying entries from the oplog for up to 24 hours without becoming stale. However, most replica sets have much lower operation volumes, and their oplogs can hold a much larger number of operations. The following factors affect how MongoDB uses space in the oplog: Update operations that affect multiple documents at once. The oplog must translate multi-updates into individual operations, in order to maintain idempotency. This can use a great deal of oplog space without a corresponding increase in disk utilization. If you delete roughly the same amount of data as you insert. In this situation the database will not grow signicantly in disk utilization, but the size of the operation log can be quite large.
392
If a signicant portion of your workload entails in-place updates. In-place updates create a large number of operations but do not change the quantity data on disk. If you can predict your replica sets workload to resemble one of the above patterns, then you may want to consider creating an oplog that is larger than the default. Conversely, if the predominance of activity of your MongoDB-based application are reads and you are writing a small amount of data, you may nd that you need a much smaller oplog. To view oplog status, including the size and the time range of operations, issue the db.printReplicationInfo() (page 983) method. For more information on oplog status, see Check the Size of the Oplog (page 446). For additional information about oplog behavior, see Oplog Internals (page 408) and Syncing (page 411). Replica Set Deployment Without replication, a standalone MongoDB instance represents a single point of failure and any disruption of the MongoDB system will render the database unusable and potentially unrecoverable. Replication increase the reliability of the database instance, and replica sets are capable of distributing reads to secondary members depending on read preference. For database work loads dominated by read operations, (i.e. read heavy) replica sets can greatly increase the capability of the database system. The minimum requirements for a replica set include two members with data, for a primary and a secondary, and an arbiter (page 388). In most circumstances, however, you will want to deploy three data members. For those deployments that rely heavily on distributing reads to secondary instances, add additional members to the set as load increases. As your deployment grows, consider adding or moving replica set members to secondary data centers or to geographically distinct locations for additional redundancy. While many architectures are possible, always ensure that the quorum of members required to elect a primary remains in your main facility. Depending on your operational requirements, you may consider adding members congured for a specic purpose including, a delayed member to help provide protection against human errors and change control, a hidden member to provide an isolated member for reporting and monitoring, and/or a secondary only member (page 388) for dedicated backups. The process of establishing a new replica set member can be resource intensive on existing members. As a result, deploy new members to existing replica sets signicantly before current demand saturates the existing members. Note: Journaling, provides single-instance write durability. The journaling greatly improves the reliability and durability of a database. Unless MongoDB runs with journaling, when a MongoDB instance terminates ungracefully, the database can end in a corrupt and unrecoverable state. You should assume that a database, running without journaling, that suffers a crash or unclean shutdown is in corrupt or inconsistent state. Use journaling, however, do not forego proper replication because of journaling. 64-bit versions of MongoDB after version 2.0 have journaling enabled by default.
Security In most cases, replica set administrators do not have to keep additional considerations in mind beyond the normal security precautions that all MongoDB administrators must take. However, ensure that: Your network conguration will allow every member of the replica set to contact every other member of the replica set.
37.1. Replica Set Fundamental Concepts
393
If you use MongoDBs authentication system to limit access to your infrastructure, ensure that you congure a keyFile (page 1081) on all members to permit authentication. For most instances, the most effective ways to control access and to secure the connection between members of a replica set depend on network-level access control. Use your environments rewall and network routing to ensure that trafc only from clients and other replica set members can reach your mongod (page 1021) instances. If needed, use virtual private networks (VPNs) to ensure secure connections over wide area networks (WANs.) Additionally, MongoDB provides an authentication mechanism for mongod (page 1021) and mongos (page 1032) instances connecting to replica sets. These instances enable authentication but specify a shared key le that serves as a shared password. New in version 1.8: Added support authentication in replica set deployments. Changed in version 1.9.1: Added support authentication in sharded replica set deployments. To enable authentication add the following option to your conguration le:
Note: You may chose to set these run-time conguration options using the --keyFile (page 1023) (or mongos --keyFile (page 1034)) options on the command line. Setting keyFile (page 1081) enables authentication and species a key le for the replica set members to use when authenticating to each other. The content of the key le is arbitrary but must be the same on all members of the replica set and on all mongos (page 1032) instances that connect to the set. The key le must be less one kilobyte in size and may only contain characters in the base64 set. The key le must not have group or world permissions on UNIX systems. Use the following command to use the OpenSSL package to generate random content for use in a key le:
Note: Key le permissions are not checked on Windows systems.
Architectures The architecture and design of the replica set deployment can have a great impact on the sets capacity and capability. This section provides a general overview of the architectural possibilities for replica set deployments. However, for most production deployments a conventional 3-member replica set with priority (page 464) values of 1 are sufcient. While the additional exibility discussed is below helpful for managing a variety of operational complexities, it always makes sense to let those complex requirements dictate complex architectures, rather than add unnecessary complexity to your deployment. Consider the following factors when developing an architecture for your replica set: Ensure that the members of the replica set will always be able to elect a primary. Run an odd number of members or run an arbiter on one of your application servers if you have an even number of members. With geographically distributed members, know where the quorum of members will be in the case of any network partitions. Attempt to ensure that the set can elect a primary among the members in the primary data center. Consider including a hidden (page 388) or delayed member (page 388) in your replica set to support dedicated functionality, like backups, reporting, and testing.
394
Consider keeping one or two members of the set in an off-site data center, but make sure to congure the priority (page 390) to prevent it from becoming primary. Create custom write concerns with replica set tags (page 455) to ensure that applications can control the threshold for a successful write operation. Use these write concerns to ensure that operations propagate to specic data centers or to machines of different functions before returning successfully. For more information regarding replica set conguration and deployments see Replica Set Architectures and Deployment Patterns (page 395).
37.2 Replica Set Architectures and Deployment Patterns

There is no single ideal replica set architecture for every deployment or environment. Indeed the exibility of replica sets might be their greatest strength. This document describes the most commonly used deployment patterns for replica sets. The descriptions are necessarily not mutually exclusive, and you can combine features of each architecture in your own deployment. For an overview of operational practices and background information, see the Architectures (page 394) section.
37.2.1 Three Member Sets

The minimum recommended architecture for a replica set consists of: One primary and Two secondary members, either of which can become the primary at any time. This makes failover (page 389) possible and ensures there exists two full and independent copies of the data set at all times. If the primary fails, the replica set elects another member as primary and continues replication until the primary recovers. Note: While not recommended, the minimum supported conguration for replica sets includes one primary, one secondary, and one arbiter (page 388). The arbiter requires fewer resources and lowers costs but sacrices operational exibility and redundancy. See also: Deploy a Replica Set (page 419).
37.2.2 Sets with Four or More Members

To increase redundancy or to provide additional resources for distributing secondary read operations, you can add additional members to a replica set. When adding additional members, ensure the following architectural conditions are true: The set has an odd number of voting members. If you have an even number of voting members, deploy an arbiter (page 388) to create an odd number. The set has no more than 7 voting members at a time. Members that cannot function as primaries in a failover have their priority (page 464) values set to 0. If a member cannot function as a primary because of resource or network latency constraints a priority (page 464) value of 0 prevents it from being a primary. Any member with a priority value greater than 0 is available to be a primary. 37.2. Replica Set Architectures and Deployment Patterns 395
A majority of the sets members operate in the main data center. See also: Add Members to a Replica Set (page 424).
37.2.3 Geographically Distributed Sets

A geographically distributed replica set provides data recovery should one data center fail. These sets include at least one member in a secondary data center. The member has its priority (page 464) set (page 467) to 0 to prevent the member from ever becoming primary. In many circumstances, these deployments consist of the following: One primary in the rst (i.e., primary) data center. One secondary member in the primary data center. This member can become the primary member at any time. One secondary member in a secondary data center. This member is ineligible to become primary. Set its local.system.replset.members[n].priority (page 464) to 0. If the primary is unavailable, the replica set will elect a new primary from the primary data center. If the connection between the primary and secondary data centers fails, the member in the secondary center cannot independently become the primary. If the primary data center fails, you can manually recover the data set from the secondary data center. With appropriate write concern (page 398) there will be no data loss and downtime can be minimal. When you add a secondary data center, make sure to keep an odd number of members overall to prevent ties during elections for primary by deploying an arbiter (page 388) in your primary data center. For example, if you have three members in the primary data center and add a member in a secondary center, you create an even number. To create an odd number and prevent ties, deploy an arbiter (page 388) in your primary data center. See also: Deploy a Geographically Distributed Replica Set (page 429)
37.2.4 Non-Production Members

In some cases it may be useful to maintain a member that has an always up-to-date copy of the entire data set but that cannot become primary. You might create such a member to provide backups, to support reporting operations, or to act as a cold standby. Such members fall into one or more of the following categories: Low-Priority: These members have local.system.replset.members[n].priority (page 464) settings such that they are either unable to become primary or very unlikely to become primary. In all other respects these low-priority members are identical to other replica set member. (See: Secondary-Only Members (page 388).) Hidden: These members cannot become primary and the set excludes them from the output of db.isMaster() (page 981) and from the output of the database command isMaster (page 838). Excluding hidden members from such outputs prevents clients and drivers from using hidden members for secondary reads. (See: Hidden Members (page 388).) Voting: This changes the number of votes that a member of the replica set has in elections. In general, use priority to control the outcome of elections, as weighting votes introduces operational complexities and risks. Only modify the number of votes when you need to have more than 7 members in a replica set. (See: Non-Voting Members (page 389).)
396
Note: All members of a replica set vote in elections except for non-voting (page 389) members. Priority, hidden, or delayed status does not affect a members ability to vote in an election.
Backups For some deployments, keeping a replica set member for dedicated backup purposes is operationally advantageous. Ensure this member is close, from a networking perspective, to the primary or likely primary. Ensure that the replication lag is minimal or non-existent. To create a dedicated hidden member (page 388) for the purpose of creating backups. If this member runs with journaling enabled, you can safely use standard block level backup methods (page 46) to create a backup of this member. Otherwise, if your underlying system does not support snapshots, you can connect mongodump (page 1044) to create a backup directly from the secondary member. In these cases, use the --oplog (page 1046) option to ensure a consistent point-in-time dump of the database state. See also: Backup Strategies for MongoDB Systems (page 41). Delayed Replication Delayed members are special mongod (page 1021) instances in a replica set that apply operations from the oplog on a delay to provide a running historical snapshot of the data set, or a rolling backup. Typically these members provide protection against human error, such as unintentionally deleted databases and collections or failed application upgrades or migrations. Otherwise, delayed member function identically to secondary members, with the following operational differences: they are not eligible for election to primary and do not receive secondary queries. Delayed members do vote in elections for primary. See Replica Set Delayed Nodes (page 388) for more information about conguring delayed replica set members. Reporting Typically hidden members provide a substrate for reporting purposes, because the replica set segregates these instances from the cluster. Since no secondary reads reach hidden members, they receive no trafc beyond what replication requires. While hidden members are not electable as primary, they are still able to vote in elections for primary. If your operational parameters requires this kind of reporting functionality, see Hidden Replica Set Nodes (page 388) and hidden (page 464) for more information regarding this functionality. Cold Standbys For some sets, it may not be possible to initialize a new member in a reasonable amount of time. In these situations, it may be useful to maintain a secondary member with an up-to-date copy for the purpose of replacing another member in the replica set. In most cases, these members can be ordinary members of the replica set, but in large sets, with varied hardware availability, or given some patterns of geographical distribution (page 396), you may want to use a member with a different priority, hidden, or voting status. Cold standbys may be valuable when your primary and hot standby secondaries members have a different hardware specication or connect via a different network than the main set. In these cases, deploy members with priority equal to 0 to ensure that they will never become primary. These members will vote in elections for primary but will never be eligible for election to primary. Consider likely failover scenarios, such as inter-site network partitions, and ensure there will be members eligible for election as primary and a quorum of voting members in the main facility. 37.2. Replica Set Architectures and Deployment Patterns 397
Note: If your set already has 7 members, set the local.system.replset.members[n].votes (page 465) value to 0 for these members, so that they wont vote in elections. See also: Secondary Only (page 388), and Hidden Nodes (page 388).
37.2.5 Arbiters
Deploy an arbiter to ensure that a replica set will have a sufcient number of members to elect a primary. While having replica sets with 2 members is not recommended for production environments, if you have just two members, deploy an arbiter. Also, for any replica set with an even number of members, deploy an arbiter. To deploy an arbiter, see the Add an Arbiter to Replica Set (page 448).
37.3 Replica Set Considerations and Behaviors for Applications and Development
From the perspective of a client application, whether a MongoDB instance is running as a single server (i.e. standalone) or a replica set is transparent. However, replica sets offer some conguration options for write and read operations. 2
37.3.1 Write Concern

Write concern is a quality of every write operation issued to a MongoDB deployment, and describes the amount of concern the application has for the outcome of the write operation. With weak or disabled write concern, the application can send an write operation to MongoDB and then continue without waiting for a response from the database. With stronger write concerns, write operations wait until MongoDB acknowledges or conrms a successful write operation. MongoDB provides different levels of write concern to better address the specic needs of applications. See also: Write Concern Reference (page 183) for a reference of specic write concern conguration. Also consider Write Operations (page 179) for a general overview of write operations with MongoDB. Note: After the driver write concern change (page 1185) all ofcially supported MongoDB drivers enable write concern by default.
Types of Write Concern Clients issue write operations with some level of write concern, which describes the level of concern or guarantee the server will provide in its response to a write operation. Consider the following levels of conceptual write concern, listed from weakest to strongest:
2
Sharded clusters where the shards are also replica sets provide the same conguration options with regards to write and read operations.
398
Errors Ignored
With a errors ignored write concern, write operations are not acknowledged by MongoDB. Write operations may not succeed in the case of connection errors that the client is not yet aware of, or if the mongod (page 1021) produces an exception (e.g. a duplicate key exception for unique indexes (page 334).) While this operation is efcient because it does not require the database to respond to every write operation, it also incurs a signicant risk with regards to the persistence and durability of the data. Warning: Do not use this option in normal operation.
Unacknowledged
With a unacknowledged write concern, MongoDB does not acknowledge the receipt of write operation as with a write concern level of ignore; however, the driver will receive and handle network errors, as possible given system networking conguration. Before the releases outlined in Default Write Concern Change (page 1185), this was the default write concern.
Acknowledged
With a receipt acknowledged write concern, the mongod (page 1021) conrms the receipt of the write operation. With this level of write concern, clients can catch network, duplicate key, and other exceptions. This is the default write concern, after the releases outlined in Default Write Concern Change (page 1185). Internally, the default write concern is to call getLastError (page 831) with no arguments. For replica sets, you can dene the default write concern settings in the getLastErrorDefaults (page 466) If getLastErrorDefaults (page 466) does not dene a default write concern setting, getLastError (page 831) defaults to basic receipt acknowledgment.
Journaled
With a journaled write concern, the mongod (page 1021) will conrm the write operation only after it has written the operation to the journal. This conrms that the write operation can survive a mongod (page 1021) shutdown and ensures that the write operation is durable. While receipt acknowledged without journaled provides the fundamental basis for write concern, there is a window between journal commits where the write operation is not fully durable. See journalCommitInterval (page 1083) for more information on this window. Require journaled as part of the write concern to provide this durability guarantee.
Replica Acknowledged
Replica sets present an additional layer of consideration for write concern. Basic write concern levels affect the write operation on only one mongod (page 1021) instance. The w argument to getLastError (page 831) provides a replica acknowledged level of write concern. With replica acknowledged you can guarantee that the write operation has propagated to the members of a replica set. See the Write Concern for Replica Sets (page 399) document for more information. Note: Requiring journaled write concern in a replica set only requires a journal commit of the write operation to the primary of the set regardless of the level of replica acknowledged write concern.
37.3. Replica Set Considerations and Behaviors for Applications and Development
399
Write Concern for Replica Sets MongoDBs built-in write concern conrms the success of write operations to a replica sets primary. Write concern uses the getLastError (page 831) command after write operations to return an object with error information or conrmation that there are no errors. From the perspective of a client application, whether a MongoDB instance is running as a single server (i.e. standalone) or a replica set is transparent. However, replica sets offer some conguration options for write and read operations. 3
Verify Write Operations
The default write concern conrms write operations only on the primary. You can congure write concern to conrm write operations to additional replica set members as well by issuing the getLastError (page 831) command with the w option. The w option conrms that write operations have replicated to the specied number of replica set members, including the primary. You can either specify a number or specify majority, which ensures the write propagates to a majority of set members. If you specify a w value greater than the number of members that hold a copy of the data (i.e., greater than the number of non-arbiter members), the operation blocks until those members become available. This can cause the operation to block forever. To specify a timeout threshold for the getLastError (page 831) operation, use the wtimeout argument. See getLastError Examples (page 831) for example invocations.
Modify Default Write Concern
You can congure your own default getLastError (page 831) behavior for a replica set. Use the getLastErrorDefaults (page 466) setting in the replica set conguration (page 463). The following sequence of commands creates a conguration that waits for the write operation to complete on a majority of the set members before returning:
cfg = rs.conf() cfg.settings = {} cfg.settings.getLastErrorDefaults = {w: "majority"} rs.reconfig(cfg)
The getLastErrorDefaults (page 466) setting affects only those getLastError (page 831) commands that have no other arguments. Note: Use of insufcient write concern can lead to rollbacks (page 390) in the case of replica set failover (page 389). Always ensure that your operations have specied the required write concern for your application. See also: Write Concern (page 180) and Write Concern Options (page 1110)
3
Sharded clusters where the shards are also replica sets provide the same conguration options with regards to write and read operations.
400
Custom Write Concerns
You can use replica set tags to create custom write concerns using the getLastErrorDefaults (page 466) and getLastErrorModes (page 466) replica set settings. Note: Custom write concern modes specify the eld name and a number of distinct values for that eld. By contrast, read preferences use the value of elds in the tag document to direct read operations. In some cases, you may be able to use the same tags for read preferences and write concerns; however, you may need to create additional tags for write concerns depending on the requirements of your application.
Single Tag Write Concerns
Consider a ve member replica set, where each member has one of the following tag sets:
{ { { { { "use": "use": "use": "use": "use": "reporting" } "backup" } "application" } "application" } "application" }
You could create a custom write concern mode that will ensure that applicable write operations will not return until members with two different values of the use tag have acknowledged the write operation. Create the mode with the following sequence of operations in the mongo (page 1036) shell:
cfg = rs.conf() cfg.settings = { getLastErrorModes: { use2: { "use": 2 } } } rs.reconfig(cfg)
To use this mode pass the string multiUse to the w option of getLastError (page 831) as follows:
db.runCommand( { getLastError: 1, w: use2 } )
Specic Custom Write Concerns
If you have a three member replica with the following tag sets:
{ "disk": "ssd" } { "disk": "san" } { "disk": "spinning" }
You cannot specify a custom getLastErrorModes (page 466) value to ensure that the write propagates to the san before returning. However, you may implement this write concern policy by creating the following additional tags, so that the set resembles the following:
{ "disk": "ssd" } { "disk": "san", "disk.san": "san" } { "disk": "spinning" }
Then, create a custom getLastErrorModes (page 466) value, as follows:

cfg = rs.conf() cfg.settings = { getLastErrorModes: { san: { "disk.san": 1 } } } rs.reconfig(cfg)
To use this mode pass the string san to the w option of getLastError (page 831) as follows:
401
db.runCommand( { getLastError: 1, w: san } )
This operation will not return until a replica set member with the tag disk.san returns. You may set a custom write concern mode as the default write concern mode using getLastErrorDefaults (page 466) replica set as in the following setting:
cfg = rs.conf() cfg.settings.getLastErrorDefaults = { ssd:1 } rs.reconfig(cfg)
See also: Congure Replica Set Tag Sets (page 455) for further information about replica set reconguration and tag sets.
37.3.2 Read Preference

Read preference describes how MongoDB clients route read operations to members of a replica set. Background By default, an application directs its read operations to the primary member in a replica set. Reading from the primary guarantees that read operations reect the latest version of a document. However, for an application that does not require fully up-to-date data, you can improve read throughput, or reduce latency, by distributing some or all reads to secondary members of the replica set. The following are use cases where you might use secondary reads: Running systems operations that do not affect the front-end application, operations such as backups and reports. Providing low-latency queries for geographically distributed deployments. If one secondary is closer to an application server than the primary, you may see better performance for that application if you use secondary reads. Providing graceful degradation in failover (page 389) situations where a set has no primary for 10 seconds or more. In this use case, you should give the application the primaryPreferred (page 403) read preference, which prevents the application from performing reads if the set has no primary. MongoDB drivers allow client applications to congure a read preference on a per-connection, per-collection, or per-operation basis. For more information about secondary read operations in the mongo (page 1036) shell, see the readPref() method. For more information about a drivers read preference conguration, see the appropriate MongoDB Drivers and Client Libraries (page 555) API documentation. Note: Read preferences affect how an application selects which member to use for read operations. As a result read preferences dictate if the application receives stale or current data from MongoDB. Use appropriate write concern policies to ensure proper data replication and consistency. If read operations account for a large percentage of your applications trafc, distributing reads to secondary members can improve read throughput. However, in most cases sharding (page 485) provides better support for larger scale operations, as clusters can distribute read and write operations across a group of machines.
Read Preference Modes
New in version 2.2. MongoDB drivers (page 555) support ve read preference modes: 402 Chapter 37. Replica Set Use and Operation
primary (page 403) primaryPreferred (page 403) secondary (page 403) secondaryPreferred (page 404) nearest (page 404) You can specify a read preference mode on connection objects, database object, collection object, or per-operation. The syntax for specifying the read preference mode is specic to the driver and to the idioms of the host language. Read preference modes are also available to clients connecting to a sharded cluster through a mongos (page 1032). The mongos (page 1032) instance obeys specied read preferences when connecting to the replica set that provides each shard in the cluster. In the mongo (page 1036) shell, the readPref() (page 963) cursor method provides access to read preferences. Warning: All read preference modes except primary (page 403) may return stale data as secondaries replicate operations from the primary with some delay. Ensure that your application can tolerate stale data if you choose to use a non-primary (page 403) mode. For more information, see read preference background (page 402) and read preference behavior (page 405). See also the documentation for your driver. primary All read operations use only the current replica set primary. This is the default. If the primary is unavailable, read operations produce an error or throw an exception. The primary (page 403) read preference mode is not compatible with read preference modes that use tag sets (page 404). If you specify a tag set with primary (page 403), the driver will produce an error. primaryPreferred In most situations, operations read from the primary member of the set. However, if the primary is unavailable, as is the case during failover situations, operations read from secondary members. When the read preference includes a tag set (page 404), the client reads rst from the primary, if available, and then from secondaries that match the specied tags. If no secondaries have matching tags, the read operation produces an error. Since the application may receive data from a secondary, read operations using the primaryPreferred (page 403) mode may return stale data in some situations. Warning: Changed in version 2.2: mongos (page 1032) added full support for read preferences. When connecting to a mongos (page 1032) instance older than 2.2, using a client that supports read preference modes, primaryPreferred (page 403) will send queries to secondaries. secondary Operations read only from the secondary members of the set. If no secondaries are available, then this read operation produces an error or exception. Most sets have at least one secondary, but there are situations where there may be no available secondary. For example, a set with a primary, a secondary, and an arbiter may not have any secondaries if a member is in recovering state or unavailable. When the read preference includes a tag set (page 404), the client attempts to nd secondary members that match the specied tag set and directs reads to a random secondary from among the nearest group (page 406).
403
If no secondaries have matching tags, the read operation produces an error.
Read operations using the secondary (page 403) mode may return stale data. secondaryPreferred In most situations, operations read from secondary members, but in situations where the set consists of a single primary (and no other members,) the read operation will use the sets primary. When the read preference includes a tag set (page 404), the client attempts to nd a secondary member that matches the specied tag set and directs reads to a random secondary from among the nearest group (page 406). If no secondaries have matching tags, the read operation produces an error. Read operations using the secondaryPreferred (page 404) mode may return stale data. nearest The driver reads from the nearest member of the set according to the member selection (page 406) process. Reads in the nearest (page 404) mode do not consider the members type. Reads in nearest (page 404) mode may read from both primaries and secondaries. Set this mode to minimize the effect of network latency on read operations without preference for current or stale data. If you specify a tag set (page 404), the client attempts to nd a replica set member that matches the specied tag set and directs reads to an arbitrary member from among the nearest group (page 406). Read operations using the nearest (page 404) mode may return stale data. Note: All operations read from a member of the nearest group of the replica set that matches the specied read preference mode. The nearest (page 404) mode prefers low latency reads over a members primary or secondary status. For nearest (page 404), the client assembles a list of acceptable hosts based on tag set and then narrows that list to the host with the shortest ping time and all other members of the set that are within the local threshold, or acceptable latency. See Member Selection (page 406) for more information.
Tag Sets
Tag sets allow you to specify custom read preferences (page 402) and write concerns (page 180) so that your application can target operations to specic members, based on custom parameters. Note: Consider the following properties of read preferences: Custom read preferences and write concerns evaluate tags sets in different ways. Read preferences consider the value of a tag when selecting a member to read from. Write concerns ignore the value of a tag to when selecting a member except to consider whether or not the value is unique. You can specify tag sets with the following read preference modes: primaryPreferred (page 403) secondary (page 403) secondaryPreferred (page 404)
4 If your set has more than one secondary, and you use the secondary (page 403) read preference mode, consider the following effect. If you have a three member replica set (page 395) with a primary and two secondaries, and if one secondary becomes unavailable, all secondary (page 403) queries must target the remaining secondary. This will double the load on this secondary. Plan and provide capacity to support this as needed.
404
nearest (page 404) You cannot specify tag sets with the primary (page 403) read preference mode. Tags are not compatible with primary (page 403) and only apply when selecting (page 406) a secondary member of a set for a read operation. However, the nearest (page 404) read mode, when combined with a tag set will select the nearest member that matches the specied tag set, which may be a primary or secondary. All interfaces use the same member selection logic (page 406) to choose the member to which to direct read operations, basing the choice on read preference mode and tag sets. For information on conguring tag sets, see Congure Replica Set Tag Sets (page 455). For more information on how read preference modes (page 402) interact with tag sets, see the documentation for each read preference mode.
Behavior
Changed in version 2.2. Auto-Retry Connection between MongoDB drivers and mongod (page 1021) instances in a replica set must balance two concerns: 1. The client should attempt to prefer current results, and any connection should read from the same member of the replica set as much as possible. 2. The client should minimize the amount of time that the database is inaccessible as the result of a connection issue, networking problem, or failover in a replica set. As a result, MongoDB drivers and mongos (page 1032): Reuse a connection to specic mongod (page 1021) for as long as possible after establishing a connection to that instance. This connection is pinned to this mongod (page 1021). Attempt to reconnect to a new member, obeying existing read preference modes (page 402), if the connection to mongod (page 1021) is lost. Reconnections are transparent to the application itself. If the connection permits reads from secondary members, after reconnecting, the application can receive two sequential reads returning from different secondaries. Depending on the state of the individual secondary members replication, the documents can reect the state of your database at different moments. Return an error only after attempting to connect to three members of the set that match the read preference mode (page 402) and tag set (page 404). If there are fewer than three members of the set, the client will error after connecting to all existing members of the set. After this error, the driver selects a new member using the specied read preference mode. In the absence of a specied read preference, the driver uses primary (page 403). After detecting a failover situation, possible.
5
the driver attempts to refresh the state of the replica set as quickly as
5 When a failover occurs, all members of the set close all client connections that produce a socket error in the driver. This behavior prevents or minimizes rollback.
405
Request Association Reads from secondary may reect the state of the data set at different points in time because secondary members of a replica set may lag behind the current state of the primary by different amounts. To prevent subsequent reads from jumping around in time, the driver can associate application threads to a specic member of the set after the rst read. The thread will continue to read from the same member until: The application performs a read with a different read preference. The thread terminates. The client receives a socket exception, as is the case when theres a network error or when the mongod (page 1021) closes connections during a failover. This triggers a retry (page 405), which may be transparent to the application. If an application thread issues a query with the primaryPreferred (page 403) mode while the primary is inaccessible, the thread will carry the association with that secondary for the lifetime of the thread. The thread will associate with the primary, if available, only after issuing a query with a different read preference, even if a primary becomes available. By extension, if a thread issues a read with the secondaryPreferred (page 404) when all secondaries are down, it will carry an association with the primary. This application thread will continue to read from the primary even if a secondary becomes available later in the threads lifetime. Member Selection Clients, by way of their drivers, and mongos (page 1032) instances for sharded clusters periodically update their view of the replica sets state: which members are up or down, which member is primary, and the latency to each mongod (page 1021) instance. For any operation that targets a member other than the primary, the driver: 1. Assembles a list of suitable members, taking into account member type (i.e. secondary, primary, or all members.) 2. Excludes members not matching the tag sets, if specied. 3. Determines which suitable member is the closest to the client in absolute terms. 4. Builds a list of members that are within a dened ping distance (in milliseconds) of the absolute nearest member. 6 5. Selects a member from these hosts at random. The member receives the read operation. Once the application selects a member of the set to use for read operations, the driver continues to use this connection for read preference until the application species a new read preference or something interrupts the connection. See Request Association (page 406) for more information. Sharding and mongos Changed in version 2.2: Before version 2.2, mongos (page 1032) did not support the read preference mode semantics (page 402). In most sharded clusters, a replica set provides each shard where read preferences are also applicable. Read operations in a sharded cluster, with regard to read preference, are identical to unsharded replica sets. Unlike simple replica sets, in sharded clusters, all interactions with the shards pass from the clients to the mongos (page 1032) instances that are actually connected to the set members. mongos (page 1032) is responsible for the application of the read preferences, which is transparent to applications.
6 Applications can congure the threshold used in this stage. The default acceptable latency is 15 milliseconds, which you can override in the drivers with their own secondaryAcceptableLatencyMS option. For mongos (page 1032) you can use the --localThreshold (page 1035) or localThreshold (page 1090) runtime options to set this value.
406
There are no conguration changes required for full support of read preference modes in sharded environments, as long as the mongos (page 1032) is at least version 2.2. All mongos (page 1032) maintain their own connection pool to the replica set members. As a result: A request without a specied preference has primary (page 403), the default, unless, the mongos (page 1032) reuses an existing connection that has a different mode set. Always explicitly set your read preference mode to prevent confusion. All nearest (page 404) and latency calculations reect the connection between the mongos (page 1032) and the mongod (page 1021) instances, not the client and the mongod (page 1021) instances. This produces the desired result, because all results must pass through the mongos (page 1032) before returning to the client. Database Commands Because some database commands read and return data from the database, all of the ofcial drivers support full read preference mode semantics (page 402) for the following commands: group (page 810) mapReduce (page 814) 7 aggregate (page 808) collStats (page 873) dbStats (page 877) count (page 808) distinct (page 809) geoNear (page 822) geoSearch (page 823) geoWalk (page 823) mongos (page 1032) currently does not route commands using read preferences; clients send all commands to shards primaries. See SERVER-7423.
Uses for non-Primary Read Preferences
You must exercise care when specifying read preferences: modes other than primary (page 403) can and will return stale data. These secondary queries will not include the most recent write operations to the replica sets primary. Nevertheless, there are several common use cases for using non-primary (page 403) read preference modes: Reporting and analytics workloads. Having these queries target a secondary helps distribute load and prevent these operations from affecting the main workload of the primary. Also consider using secondary (page 403) in conjunction with a direct connection to a hidden member (page 388) of the set.
7 Only inline mapReduce (page 814) operations that do not write data support read preference, otherwise these operations must run on the primary members.
407
Providing local reads for geographically distributed applications. If you have application servers in multiple data centers, you may consider having a geographically distributed replica set (page 396) and using a non primary read preference or the nearest (page 404) to avoid network latency. Maintaining availability during a failover. Use primaryPreferred (page 403) if you want your application to do consistent reads from the primary under normal circumstances, but to allow stale reads from secondaries in an emergency. This provides a readonly mode for your application during a failover. Warning: In some situations using secondaryPreferred (page 404) to distribute read load to replica sets may carry signicant operational risk: if all secondaries are unavailable and your set has enough arbiters to prevent the primary from stepping down, then the primary will receive all trafc from clients. For this reason, use secondary (page 403) to distribute read load to replica sets, not secondaryPreferred (page 404). Using read modes other than primary (page 403) and primaryPreferred (page 403) to provide extra capacity is not in and of itself justication for non-primary (page 403) in many cases. Furthermore, sharding (page 483) increases read and write capacity by distributing read and write operations across a group of machines.
37.4 Replica Set Internals and Behaviors

This document provides a more in-depth explanation of the internals and operation of replica set features. This material is not necessary for normal operation or application development but may be useful for troubleshooting and for further understanding MongoDBs behavior and approach. For additional information about the internals of replication replica sets see the following resources in the MongoDB Manual: The local Database (page 1097) Replica Set Commands (page 468) db.getReplicationInfo() (page 979) Replica Set Conguration (page 463)
37.4.1 Oplog Internals

For an explanation of the oplog, see Oplog (page 392). Under various exceptional situations, updates to a secondarys oplog might lag behind the desired performance time. See Replication Lag (page 444) for details. All members of a replica set send heartbeats (pings) to all other members in the set and can import operations to the local oplog from any other member in the set. Replica set oplog operations are idempotent. The following operations require idempotency: initial sync post-rollback catch-up sharding chunk migrations
408
37.4.2 Read Preference Internals

MongoDB uses single-master replication to ensure that the database remains consistent. However, clients may modify the read preferences (page 402) on a per-connection basis in order to distribute read operations to the secondary members of a replica set. Read-heavy deployments may achieve greater query throughput by distributing reads to secondary members. But keep in mind that replication is asynchronous; therefore, reads from secondaries may not always reect the latest writes to the primary. See also: Consistency (page 390) Note: Use the output from db.getReplicationInfo() (page 979) run on a secondary member to asses the current state of replication and determine if there is any unintended replication delay.
37.4.3 Member Congurations

Replica sets can include members with the following four special congurations that affect membership behavior: Secondary-only (page 388) members have their priority (page 464) values set to 0 and thus are not eligible for election as primaries. Hidden (page 388) members do not appear in the output of db.isMaster() (page 981). This prevents clients from discovering and potentially querying the member in question. Delayed (page 388) members lag a xed period of time behind the primary. These members are typically used for disaster recovery scenarios. For example, if an administrator mistakenly truncates a collection, and you discover the mistake within the lag window, then you can manually fail over to the delayed member. Arbiters (page 388) exist solely to participate in elections. They do not replicate data from the primary. In almost every case, replica sets simplify the process of administering database replication. However, replica sets still have a unique set of administrative requirements and concerns. Choosing the right system architecture (page 395) for your data set is crucial. See also: Member Congurations (page 387)
37.4.4 Security Internals

Administrators of replica sets also have unique monitoring (page 97) and security (page 393) concerns. The replica set functions (page 987) in the mongo (page 1036) shell, provide the tools necessary for replica set administration. In particular use the rs.conf() (page 988) to return a document that holds the replica set conguration (page 463) and use rs.reconfig() (page 989) to modify the conguration of an existing replica set.
37.4.5 Election Internals

Elections are the process replica set members use to select which member should become primary. A primary is the only member in the replica set that can accept write operations, including insert() (page 936), update() (page 948), and remove() (page 944). The following events can trigger an election: You initialize a replica set for the rst time.
37.4. Replica Set Internals and Behaviors
409
A primary steps down. A primary will step down in response to the replSetStepDown (page 844) command or if it sees that one of the current secondaries is eligible for election and has a higher priority. A primary also will step down when it cannot contact a majority of the members of the replica set. When the current primary steps down, it closes all open client connections to prevent clients from unknowingly writing data to a non-primary member. A secondary member loses contact with a primary. A secondary will call for an election if it cannot establish a connection to a primary. A failover occurs. In an election, all members have one vote, including hidden (page 388) members, arbiters (page 388), and even recovering members. Any mongod (page 1021) can veto an election. In the default conguration, all members have an equal chance of becoming primary; however, its possible to set priority (page 464) values that weight the election. In some architectures, there may be operational reasons for increasing the likelihood of a specic replica set member becoming primary. For instance, a member located in a remote data center should not become primary. See: Member Priority (page 390) for more information. Any member of a replica set can veto an election, even if the member is a non-voting member (page 389). A member of the set will veto an election under the following conditions: If the member seeking an election is not a member of the voters set. If the member seeking an election is not up-to-date with the most recent operation accessible in the replica set. If the member seeking an election has a lower priority than another member in the set that is also eligible for election. If a secondary only member (page 388) 8 is the most current member at the time of the election, another eligible member of the set will catch up to the state of this secondary member and then attempt to become primary. If the current primary member has more recent operations (i.e. a higher optime) than the member seeking election, from the perspective of the voting member. The current primary will veto an election if it has the same or more recent operations (i.e. a higher or equal optime) than the member seeking election. The rst member to receive votes from a majority of members in a set becomes the next primary until the next election. Be aware of the following conditions and possible situations: Replica set members send heartbeats (pings) to each other every 2 seconds. If a heartbeat does not return for more than 10 seconds, the other members mark the delinquent member as inaccessible. Replica set members compare priorities only with other members of the set. The absolute value of priorities does not have any impact on the outcome of replica set elections, with the exception of the value 0, which indicates the member cannot become primary and cannot seek election. For details, see Congure a Non-Voting Replica Set Member (page 452). A replica set member cannot become primary unless it has the highest optime of any visible member in the set. If the member of the set with the highest priority is within 10 seconds of the latest oplog entry, then the set will not elect a primary until the member with the highest priority catches up to the latest operation. A replica set member cannot become primary unless it can connect to a majority of the replica set. For the purposes of elections, a majority refers to the total number of votes, not number of members. Do not alter the number of votes in a replica set, except in exceptional cases. See also:
8
Remember that hidden (page 388) and delayed (page 388) imply secondary-only (page 388) conguration.
410
Non-voting members in a replica set (page 389), Congure a Non-Voting Replica Set Member (page 452), and replica configuration (page 465).
37.4.6 Syncing
In order to remain up-to-date with the current state of the replica set, set members sync, or copy, oplog entries from other members. Members sync data at two different points: Initial sync occurs when MongoDB creates new databases on a new or restored member, populating the member with the replica sets data. When a new or restored member joins or rejoins a set, the member waits to receive heartbeats from other members. By default, the member syncs from the closest member of the set that is either the primary or another secondary with more recent oplog entries. This prevents two secondaries from syncing from each other. Replication occurs continually after initial sync and keeps the member updated with changes to the replica sets data. In MongoDB 2.0, secondaries only change sync targets if the connection to the sync target drops error. For example: 1. If you have two secondary members in one data center and a primary in a second facility, and if you start all three instances at roughly the same time (i.e. with no existing data sets or oplog), both secondaries will likely sync from the primary, as neither secondary has more recent oplog entries. If you restart one of the secondaries, then when it rejoins the set it will likely begin syncing from the other secondary, because of proximity. 2. If you have a primary in one facility and a secondary in an alternate facility, and if you add another secondary to the alternate facility, the new secondary will likely sync from the existing secondary because it is closer than the primary. In MongoDB 2.2, secondaries also use the following additional sync behaviors: Secondaries will sync from delayed members (page 388) only if no other member is available. Secondaries will not sync from hidden members (page 388). Secondaries will not start syncing from a member in a recovering state. For one member to sync from another, both members must have the same value, either true or false, for the buildIndexes (page 464) eld.
9
or produces an
37.4.7 Multithreaded Replication

MongoDB applies write operations in batches using a multithreaded approach. The replication process divides each batch among a group of threads which apply many operations with greater concurrency. Even though threads may apply operations out of order, a client reading data from a secondary will never return documents that reect an in-between state that never existed on the primary. To ensure this consistency, MongoDB blocks all read operations while applying the batch of operations. To help improve the performance of operation application, MongoDB fetches all the memory pages that hold data and indexes that the operations in the batch will affect. The prefetch stage minimizes the amount of time MongoDB must hold the write lock to apply operations. See the replIndexPrefetch (page 1088) setting to modify the index fetching behavior.
Secondaries will stop syncing from a member if the connection used to poll oplog entries is unresponsive for 30 seconds. If a connection times out, the member may select a new member to sync from.
9
37.4. Replica Set Internals and Behaviors
411
37.4.8 Pre-Fetching Indexes to Improve Replication Throughput

By default, secondaries will in most cases pre-fetch Indexes (page 327) associated with the affected document to improve replication throughput. You can limit this feature to pre-fetch only the index on the _id eld, or you can disable this feature entirely. For more information, see replIndexPrefetch (page 1088). The following document describes master-slave replication, which is deprecated. Use replica sets instead of masterslave in all new deployments.
37.5 Master Slave Replication

Deprecated since version 1.6: Replica sets (page 387) replace master-slave replication. Use replica sets rather than master-slave replication for all new production deployments. Replica sets provide functional super-set of master-slave and are more robust for production use. Master-slave replication preceded replica and makes it possible have a large number of non-master (i.e. slave) and to only replicate operations for a single database; however, master-slave replication provides less redundancy, and does not automate failover. See Deploy Master-Slave Equivalent using Replica Sets (page 414) for a replica set conguration that is equivalent to master-slave replication. Warning: This documentation remains to support legacy deployments and for archival purposes, only.
37.5.1 Fundamental Operations

Initial Deployment To congure a master-slave deployment, start two mongod (page 1021) instances: one in master (page 1088) mode, and the other in slave (page 1088) mode. To start a mongod (page 1021) instance in master (page 1088) mode, invoke mongod (page 1021) as follows:
mongod --master --dbpath /data/masterdb/
With the --master (page 1029) option, the mongod (page 1021) will create a local.oplog.$main (page 1099) collection, which the operation log that queues operations that the slaves will apply to replicate operations from the master. The --dbpath (page 1023) is optional. To start a mongod (page 1021) instance in slave (page 1088) mode, invoke mongod (page 1021) as follows:
mongod --slave --source <masterhostname><:<port>> --dbpath /data/slavedb/
Specify the hostname and port of the master instance to the --source (page 1029) argument. The --dbpath (page 1023) is optional. For slave (page 1088) instances, MongoDB stores data about the source server in the local.sources (page 1099) collection. Conguration Options for Master-Slave Deployments As an alternative to specifying the --source (page 1029) run-time option, can add a document to local.sources (page 1099) specifying the master (page 1088) instance, as in the following operation in the mongo (page 1036) shell: 412 Chapter 37. Replica Set Use and Operation
1 2 3
use local db.sources.find() db.sources.insert( { host: <masterhostname> <,only: databasename> } );
In line 1, you switch context to the local database. In line 2, the find() (page 924) operation should return no documents, to ensure that there are no documents in the sources collection. Finally, line 3 uses db.collection.insert() (page 936) to insert the source document into the local.sources (page 1099) collection. The model of the local.sources (page 1099) document is as follows: host The host eld species the master (page 1088)mongod (page 1021) instance, and holds a resolvable hostname, i.e. IP address, or a name from a host le, or preferably a fully qualied domain name. You can append <:port> to the host name if the mongod (page 1021) is not running on the default 27017 port. only Optional. Specify a name of a database. When specied, MongoDB will only replicate the indicated database. Operational Considerations for Replication with Master Slave Deployments Master instances store operations in an oplog which is a capped collection (page 558). As a result, if a slave falls too far behind the state of the master, it cannot catchup and must re-sync from scratch. Slave may become out of sync with a master if: The slave falls far behind the data updates available from that master. The slave stops (i.e. shuts down) and restarts later after the master has overwritten the relevant operations from the master. When slaves, are out of sync, replication stops. Administrators must intervene manually to restart replication. Use the resync (page 846) command. Alternatively, the --autoresync (page 1029) allows a slave to restart replication automatically, after ten second pause, when the slave falls out of sync with the master. With --autoresync (page 1029) specied, the slave will only attempt to re-sync once in a ten minute period. To prevent these situations you should specify a larger oplog when you start the master (page 1088) instance, by adding the --oplogSize (page 1028) option when starting mongod (page 1021). If you do not specify --oplogSize (page 1028), mongod (page 1021) will allocate 5% of available disk space on start up to the oplog, with a minimum of 1GB for 64bit machines and 50MB for 32bit machines.
37.5.2 Run time Master-Slave Conguration

MongoDB provides a number of run time conguration options for mongod (page 1021) instances in master-slave deployments. You can specify these options in conguration les (page 35) or on the command-line. See documentation of the following: For master nodes: master (page 1088) slave (page 1088) For slave nodes: source (page 1088) only (page 1089) slaveDelay (page 1089)
37.5. Master Slave Replication
413
Also consider the Master-Slave Replication Command Line Options (page 1029) for related options. Diagnostics On a master instance, issue the following operation in the mongo (page 1036) shell to return replication status from the perspective of the master:
db.printReplicationInfo()
On a slave instance, use the following operation in the mongo (page 1036) shell to return the replication status from the perspective of the slave:
db.printSlaveReplicationInfo()
Use the serverStatus (page 889) as in the following operation, to return status of the replication:
db.serverStatus()
See server status repl elds (page 897) for documentation of the relevant section of output.
37.5.3 Security
When running with auth (page 1081) enabled, in master-slave deployments, you must create a user account for the local database on both mongod (page 1021) instances. Log in, and authenticate to the admin database on the slave instance, and then create the repl user on the local database, with the following operation:
use local db.addUser(repl, <replpassword>)
Once created, repeat the operation on the master instance. The slave instance rst looks for a user named repl in the local.system.users (page 160) collection. If present, the slave uses this user account to authenticate to the local database in the master instance. If the repl user does not exist, the slave instance attempts to authenticate using the rst user document in the local.system.users (page 160) collection. The local database works like the admin database: an account for local has access to the entire server. See also: Security (page 129) for more information about security in MongoDB
37.5.4 Ongoing Administration and Operation of Master-Slave Deployments

Deploy Master-Slave Equivalent using Replica Sets If you want a replication conguration that resembles master-slave replication, using replica sets replica sets, consider the following replica conguration document. In this deployment hosts <master> and <slave> 10 provide replication that is roughly equivalent to a two-instance master-slave deployment:
{ _id : setName, members : [ { _id : 0, host : "<master>", priority : 1 }, { _id : 1, host : "<slave>", priority : 0, votes : 0 }
10
In replica set congurations, the host (page 463) eld must hold a resolvable hostname.
414
] }
See Replica Set Conguration (page 463) for more information about replica set congurations. Failing over to a Slave (Promotion) To permanently failover from a unavailable or damaged master (A in the following example) to a slave (B): 1. Shut down A. 2. Stop mongod (page 1021) on B. 3. Back up and move all data les that begin with local on B from the dbpath (page 1081). Warning: caution. Removing local.* is irrevocable and cannot be undone. Perform this step with extreme
4. Restart mongod (page 1021) on B with the --master (page 1029) option. Note: This is a one time operation, and is not reversible. A cannot become a slave of B until it completes a full resync.
Inverting Master and Slave If you have a master (A) and a slave (B) and you would like to reverse their roles, follow this procedure. The procedure assumes A is healthy, up-to-date and available. If A is not healthy but the hardware is okay (power outage, server crash, etc.), skip steps 1 and 2 and in step 8 replace all of As les with Bs les in step 8. If A is not healthy and the hardware is not okay, replace A with a new machine. Also follow the instructions in the previous paragraph. To invert the master and slave in a deployment: 1. Halt writes on A using the fsync command. 2. Make sure B is up to date with the state of A. 3. Shut down B. 4. Back up and move all data les that begin with local on B from the dbpath (page 1081) to remove the existing local.sources data. Warning: caution. Removing local.* is irrevocable and cannot be undone. Perform this step with extreme
5. Start B with the --master (page 1029) option. 6. Do a write on B, which primes the oplog to provide a new sync start point. 7. Shut down B. B will now have a new set of data les that start with local. 8. Shut down A and replace all les in the dbpath (page 1081) of A that start with local with a copy of the les in the dbpath (page 1081) of B that begin with local. Considering compressing the local les from B while you copy them, as they may be quite large. 9. Start B with the --master (page 1029) option.
415
10. Start A with all the usual slave options, but include fastsync (page 1029). Creating a Slave from an Existing Masters Disk Image If you can stop write operations to the master for an indenite period, you can copy the data les from the master to the new slave and then start the slave with --fastsync (page 1029). Warning: Be careful with --fastsync (page 1029). If the data on both instances is identical, a discrepancy will exist forever. fastsync (page 1088) is a way to start a slave by starting with an existing master disk image/backup. This option declares that the administrator guarantees the image is correct and completely up-to-date with that of the master. If you have a full and complete copy of data from a master you can use this option to avoid a full synchronization upon starting the slave. Creating a Slave from an Existing Slaves Disk Image You can just copy the other slaves data le snapshot without any special options. Only take data snapshots when a mongod (page 1021) process is down or locked using db.fsyncLock() (page 977). Resyncing a Slave that is too Stale to Recover Slaves asynchronously apply write operations from the master that the slaves poll from the masters oplog. The oplog is nite in length, and if a slave is too far behind, a full resync will be necessary. To resync the slave, connect to a slave using the mongo (page 1036) and issue the resync (page 846) command:
use admin db.runCommand( { resync: 1 } )
This forces a full resync of all data (which will be very slow on a large database). You can achieve the same effect by stopping mongod (page 1021) on the slave, deleting the entire content of the dbpath (page 1081) on the slave, and restarting the mongod (page 1021). Slave Chaining Slaves cannot be chained. They must all connect to the master directly. If a slave attempts slave from another slave you will see the following line in the mongod (page 1021) long of the shell:
assertion 13051 tailable cursor requested on non capped collection ns:local.oplog.$main
Correcting a Slaves Source To change a slaves source, manually modify the slaves local.sources (page 1099) collection. Example Consider the following: If you accidentally set an incorrect hostname for the slaves source (page 1088), as in the following example:
416
mongod --slave --source prod.mississippi
You can correct this, by restarting the slave without the --slave (page 1029) and --source (page 1029) arguments:
mongod
Connect to this mongod (page 1021) instance using the mongo (page 1036) shell and update the local.sources (page 1099) collection, with the following operation sequence:
use local db.sources.update( { host : "prod.mississippi" }, { $set : { host : "prod.mississippi.example.net" } } )
Restart the slave with the correct command line arguments or with no --source (page 1029) option. After conguring local.sources (page 1099) the rst time, the --source (page 1029) will have no subsequent effect. Therefore, both of the following invocations are correct:
mongod --slave --source prod.mississippi.example.net
or
mongod --slave
The slave now polls data from the correct master. For documentation of MongoDBs operational segregation capabilities for replica set deployments see the Data Center Awareness (page 61)
417
418
CHAPTER 38
Replica Set Tutorials and Procedures
The following tutorials describe a number of common replica set maintenance and operational practices in greater detail.
38.1 Replica Set Administration

Replica sets automate most administrative tasks associated with database replication. Nevertheless, several operations related to deployment and systems management require administrator intervention. The following tutorials provide task-oriented instructions for specic administrative tasks related to replica set operation.
38.1.1 Deploy a Replica Set

This tutorial describes how to create a three-member replica set from three existing mongod (page 1021) instances. The tutorial provides two procedures: one for development and test systems; and a one for production systems. To instead deploy a replica set from a single standalone MongoDB instance, see Convert a Standalone to a Replica Set (page 423). For additional information regarding replica set deployments, see Replica Set Fundamental Concepts (page 387) and Replica Set Architectures and Deployment Patterns (page 395). Overview Three member replica sets provide enough redundancy to survive most network partitions and other system failures. Additionally, these sets have sufcient capacity for many distributed read operations. Most deployments require no additional members or conguration. Requirements Most replica sets consist of three or more mongod (page 1021) instances. 1 This tutorial describes a three member set. Production environments should have at least three distinct systems so that each system can run its own instance of mongod (page 1021). For development systems you can run all three instances of the mongod (page 1021) process on a local system or within a virtual instance. For production environments, you should maintain as much separation between members as possible. For example, when using virtual machines for production deployments, each member should live on a separate host server, served by redundant power circuits and with redundant network paths.
1 To ensure smooth elections (page 389) always design replica sets with odd numbers of members. Use Arbiters (page 388) to ensure the set has odd number of voting members and avoid tied elections.
419
Procedures These procedures assume you already have instances of MongoDB installed on the systems you will add as members of your replica set. If you have not already installed MongoDB, see the installation tutorials (page 3).
Deploy a Development or Test Replica Set
The examples in this procedure create a new replica set named rs0. 1. Before creating your replica set, verify that every member can successfully connect to every other member. The network conguration must allow all possible connections between any two members. To test connectivity, see Test Connections Between all Members (page 446). 2. Start three instances of mongod (page 1021) as members of a replica set named rs0, as described in this step. For ephemeral tests and the purposes of this guide, you may run the mongod (page 1021) instances in separate windows of GNU Screen. OS X and most Linux distributions come with screen installed by default 2 systems. (a) Create the necessary data directories by issuing a command similar to the following:
mkdir -p /srv/mongodb/rs0-0 /srv/mongodb/rs0-1 /srv/mongodb/rs0-2
(b) Issue the following commands, each in a distinct screen window:

mongod --port 27017 --dbpath /srv/mongodb/rs0-0 --replSet rs0 --smallfiles --oplogSize 128 mongod --port 27018 --dbpath /srv/mongodb/rs0-1 --replSet rs0 --smallfiles --oplogSize 128 mongod --port 27019 --dbpath /srv/mongodb/rs0-2 --replSet rs0 --smallfiles --oplogSize 128
This starts each instance as a member of a replica set named rs0, each running on a distinct port. If you are already using these ports, select different ports. The --smallfiles (page 1027) and --oplogSize reduce the disk space that each mongod (page 1021) uses for the purposes of testing. For more information on these and other conguration options, see Conguration File Options (page 1078). Important: Unless otherwise indicated, only use these invocations for test deployments. 3. Open a mongo (page 1036) shell and connect to the rst mongod (page 1021) instance, with the following command:
mongo --port 27017
4. Create a replica set conguration object in the mongo (page 1036) shell environment to use to initiate the replica set with the following sequence of operations:
rsconf = { _id: "rs0", members: [ { _id: 0, host: "<hostname>:27017" } ] }
2
GNU Screen is packaged as screen on Debian-based, Fedora/Red Hat-based, and Arch Linux.
420
Chapter 38. Replica Set Tutorials and Procedures
rs.initiate( rsconf )

rs.conf()
7. Add the second and third mongod (page 1021) instances to the replica set using the rs.add() (page 987) method. Replace <hostname> with your systems hostname in the following examples:
rs.add("<hostname>:27018") rs.add("<hostname>:27019")
After these commands return you have a fully functional replica set. New replica sets elect a primary within a few seconds. 8. Check the status of your replica set at any time with the rs.status() (page 990) operation. See also: The documentation of the following shell functions for more information: rs.initiate() (page 989) rs.conf() (page 988) rs.reconfig() (page 989) rs.add() (page 987) You may also consider the simple setup script as an example of a basic automatically congured replica set.
Deploy a Production Replica Set
Production replica sets are very similar to the development or testing deployment described above, with the following differences: Each member of the replica set resides on its own machine, and the MongoDB processes all bind to port 27017, which is the standard MongoDB port. Each member of the replica set must be accessible by way of resolvable DNS or hostnames in the following scheme: mongodb0.example.net mongodb1.example.net mongodb2.example.net Congure DNS names appropriately, or set up your systems http://docs.mongodb.org/manual/etc/hosts le to reect this conguration. You specify run-time conguration on each system in a conguration le (page 1078) stored in http://docs.mongodb.org/manual/etc/mongodb.conf or in a related location. You do not specify run-time conguration through command line options. For each MongoDB instance, use the following conguration. Set conguration values appropriate to your systems:
port = 27017 bind_ip = 10.8.0.10 dbpath = /srv/mongodb/
38.1. Replica Set Administration
421
fork = true replSet = rs0
You do not need to specify an interface with bind_ip (page 1079). However, if you do not specify an interface, MongoDB listens for connections on all available IPv4 interfaces. Modify bind_ip (page 1079) to reect a secure interface on your system that is able to access all other members of the set and on which all other members of the replica set can access the current member. The DNS or host names must point and resolve to this IP address. Congure network rules or a virtual private network (i.e. VPN) to permit this access. For more documentation on run time options used above and on additional conguration options, see Conguration File Options (page 1078). To deploy a production replica set: 1. Before creating your replica set, verify that every member can successfully connect to every other member. The network conguration must allow all possible connections between any two members. To test connectivity, see Test Connections Between all Members (page 446). 2. On each system start the mongod (page 1021) process by issuing a command similar to following:
mongo
rs.initiate()

rs.conf()
6. Add two members to the replica set by issuing a sequence of commands similar to the following:
After these commands return you have a fully functional replica set. New replica sets elect a primary within a few seconds. 7. Check the status of your replica set at any time with the rs.status() (page 990) operation. See also: The documentation of the following shell functions for more information: rs.initiate() (page 989) rs.conf() (page 988) rs.reconfig() (page 989) rs.add() (page 987)
422
38.1.2 Convert a Standalone to a Replica Set

While standalone MongoDB instances are useful for testing, development and trivial deployments, for production use, replica sets provide required robustness and disaster recovery. This tutorial describes how to convert an existing standalone instance into a three-member replica set. If youre deploying a replica set fresh, without any existing MongoDB data or instance, see Deploy a Replica Set (page 419). For more information on replica sets, their use, and administration (page 385), see: Replica Set Fundamental Concepts (page 387), Replica Set Architectures and Deployment Patterns (page 395), Add Members to a Replica Set (page 424) Add an Arbiter to Replica Set (page 448) Remove Members from Replica Set (page 427) Replace a Replica Set Member (page 427) Adjust Priority for Replica Set Member (page 428) Resync a Member of a Replica Set (page 428) Congure a Secondarys Sync Target (page 454) Congure a Delayed Replica Set Member (page 451) Congure a Replica Set Member as Hidden (page 452) Congure a Non-Voting Replica Set Member (page 452) Prevent Replica Set Member from Becoming Primary (page 453) Manage Chained Replication (page 439) Troubleshoot Replica Sets (page 444), and Replica Set Considerations and Behaviors for Applications and Development (page 398). Note: If youre converting a standalone instance into a replica set that is a shard in a sharded cluster you must change the shard host information in the cong database. While connected to a mongos (page 1032) instance with a mongo (page 1036) shell, issue a command in the following form:
db.getSiblingDB("config").shards.save( {_id: "<name>", host: "<replica-set>/<member,><member,><...>"
Replace <name> with the name of the shard, replace <replica-set> with the name of the replica set, and replace <member,><member,><> with the list of the members of the replica set. After completing this operation you must restart all mongos (page 1032) instances. When possible you should restart all components of the replica sets (i.e. all mongos (page 1032) and all shard mongod (page 1021) instances.)
Procedure This procedure assumes you have a standalone instance of MongoDB installed. If you have not already installed MongoDB, see the installation tutorials (page 3). 1. Shut down the your MongoDB instance and then restart using the --replSet (page 1028) option and the name of the replica set, which is rs0 in the example below. Use a command similar to the following:
423
mongod --port 27017 --dbpath /srv/mongodb/db0 --replSet rs0
Replace http://docs.mongodb.org/manual/srv/mongodb/db0 with the path of your dbpath (page 1081). This starts the instance as a member of a replica set named rs0. For more information on conguration options, see Conguration File Options (page 1078) and the mongod (page 1021) manual page. 2. Open a mongo (page 1036) shell and connect to the mongod (page 1021) instance. In a new system shell session, use the following command to start a mongo (page 1036) shell:
mongo
3. Use rs.initiate() (page 989) to initiate the replica set:

rs.initiate()
The set is now operational. To return the replica set conguration, call the rs.conf() (page 988) method. To check the status of the replica set, use rs.status() (page 990). 4. Now add additional replica set members. On two distinct systems, start two new standalone mongod (page 1021) instances. Then, in the mongo (page 1036) shell instance connected to the rst mongod (page 1021) instance, issue a command in the following form:
rs.add("<hostname><:port>")
Replace <hostname> and <port> with the resolvable hostname and port of the mongod (page 1021) instance you want to add to the set. Repeat this operation for each mongod (page 1021) that you want to add to the set. For more information on adding hosts to a replica set, see the Add Members to a Replica Set (page 424) document.
38.1.3 Add Members to a Replica Set

Overview This tutorial explains how to add an additional member to an existing replica set. Before adding a new member to an existing replica set, do one of the following to prepare the new members data directory: Make sure the new members data directory does not contain data. The new member will copy the data from an existing member. If the new member is in a recovering state, it must exit and become a secondary before MongoDB can copy all data as part of the replication process. This process takes time but does not require administrator intervention. Manually copy the data directory from an existing member. The new member becomes a secondary member and will catch up to the current state of the replica set after a short interval. Copying the data over manually shortens the amount of time for the new member to become current. Ensure that you can copy the data directory to the new member and begin replication within the window allowed by the oplog (page 392). If the difference in the amount of time between the most recent operation and the most recent operation to the database exceeds the length of the oplog on the existing members, then the new instance will have to perform an initial sync, which completely resynchronizes the data, as described in Resync a Member of a Replica Set (page 428). Use db.printReplicationInfo() (page 983) to check the current state of replica set members with regards to the oplog. 424 Chapter 38. Replica Set Tutorials and Procedures
For background on replication deployment patterns, see the Replica Set Architectures and Deployment Patterns (page 395) document. Requirements 1. An active replica set. 2. A new MongoDB system capable of supporting your dataset, accessible by the active replica set through the network. If neither of these conditions are satised, please use the MongoDB installation tutorial (page 3) and the Deploy a Replica Set (page 419) tutorial instead. Procedures The examples in this procedure use the following conguration: The active replica set is rs0. The new member to be added is mongodb3.example.net. The mongod (page 1021) instance default port is 27017. The mongodb.conf conguration le exists in the http://docs.mongodb.org/manual/etc directory and contains the following replica set information:
port = 27017 bind_ip = 10.8.0.10 dbpath = /srv/mongodb/db0 logpath = /var/log/mongodb.log fork = true replSet = rs0
For more information on conguration options, see Conguration File Options (page 1078).
Add a Member to an Existing Replica Set
This procedure uses the above example conguration (page 425). 1. Deploy a new mongod (page 1021) instance, specifying the name of the replica set. You can do this one of two ways: Using the mongodb.conf le. On the primary, issue a command that resembles the following:
Using command line arguments. On the primary, issue command that resembles the following:
mongod --dbpath /srv/mongodb/db0 --replSet rs0
Replace http://docs.mongodb.org/manual/srv/mongodb/db0 with the path of your dbpath (page 1081). Take note of the host name and port information for the new mongod (page 1021) instance. 38.1. Replica Set Administration 425
2. Open a mongo (page 1036) shell connected to the replica sets primary:
mongo
Note: The primary is the only member that can add or remove members from the replica set. If you do not know which member is the primary, log into any member of the replica set using mongo (page 1036) and issue the db.isMaster() (page 981) command to determine which member is in the isMaster.primary (page 839) eld. For example, on the system shell:
mongo mongodb0.example.net
Then in the mongo (page 1036) shell:

db.isMaster()
If you are not connected to the primary, disconnect from the current client and reconnect to the primary. 3. In the mongo (page 1036) shell, issue the following command to add the new member to the replica set.
rs.add("mongodb3.example.net")
Note: You can also include the port number, depending on your setup:
rs.add("mongodb3.example.net:27017")
4. Verify that the member is now part of the replica set by calling the rs.conf() (page 988) method, which displays the replica set conguration (page 463):
rs.conf()
You can use the rs.status() (page 990) method to view replica set status. For a description of the status elds, see replSetGetStatus (page 840).
Add a Member to an Existing Replica Set (Alternate Procedure)
Alternately, you can add a member to a replica set by specifying an entire conguration document with some or all of the elds in a members (page 463) sub-documents. For example:
rs.add({_id: 1, host: "mongodb3.example.net:27017", priority: 0, hidden: true})
This congures a hidden member that is accessible at mongodb3.example.net:27017. See host (page 463), priority (page 464), and hidden (page 464) for more information about these settings. When you specify a full conguration object with rs.add() (page 987), you must declare the _id eld, which is not automatically populated in this case. Production Notes In production deployments you likely want to use and congure a control script to manage this process based on this command. A member can be removed from a set and re-added later. If the removed members data is still relatively fresh, it can recover and catch up from its old data set. See the rs.add() (page 987) and rs.remove() (page 990) helpers.
426
If you have a backup or snapshot of an existing member, you can move the data les (i.e. http://docs.mongodb.org/manual/data/db or dbpath (page 1081)) to a new system and use them to quickly initiate a new member. These les must be: clean: the existing dataset must be from a consistent copy of the database from a member of the same replica set. See the Backup Strategies for MongoDB Systems (page 41) document for more information. recent: the copy must more recent than the oldest operation in the primary members oplog. The new secondary must be able to become current using operations from the primarys oplog. There is a maximum of seven voting members (page 409) in any replica set. When adding more members to a replica set that already has seven votes, you must either: add the new member as a non-voting members (page 389) or, remove votes from an existing member (page 465).
38.1.4 Remove Members from Replica Set

You may remove a member of a replica set at any time; however, for best results always shut down the mongod (page 1021) instance before removing it from a replica set. Changed in version 2.2: Before 2.2, you had to shut down the mongod (page 1021) instance before removing it. While 2.2 removes this requirement, it remains good practice. To remove a member, use the rs.remove() (page 990) method in the mongo (page 1036) shell while connected to the current primary. Issue the db.isMaster() (page 981) command when connected to any member of the set to determine the current primary. Use a command in either of the following forms to remove the member:
rs.remove("mongo2.example.net:27017") rs.remove("mongo3.example.net")
This operation disconnects the shell briey and forces a re-connection as the replica set renegotiates which member will be primary. The shell displays an error even if this command succeeds. You can re-add a removed member to a replica set at any time using the procedure for adding replica set members (page 424). Additionally, consider using the replica set reconguration procedure (page 467) to change the host (page 463) value to rename a member in a replica set directly.
38.1.5 Replace a Replica Set Member

Use this procedure to replace a member of a replica set when the hostname has changed. This procedure preserves all existing conguration for a member, except its hostname/location. You may need to replace a replica set member if you want to replace an existing system and only need to change the hostname rather than completely replace all congured options related to the previous member. Use rs.reconfig() (page 989) to change the value of the host (page 463) eld to reect the new hostname or port number. rs.reconfig() (page 989) will not change the value of _id (page 463).
cfg = rs.conf() cfg.members[0].host = "mongo2.example.net:27019" rs.reconfig(cfg)
Warning: Any replica set conguration change can trigger the current primary to step down, which forces an election (page 389). This causes the current shell session, and clients connected to this replica set, to produce an error even when the operation succeeds.
427
38.1.6 Adjust Priority for Replica Set Member

To change the value of the priority (page 464) in the replica set conguration, use the following sequence of commands in the mongo (page 1036) shell:
cfg = rs.conf() cfg.members[0].priority = 0.5 cfg.members[1].priority = 2 cfg.members[2].priority = 2 rs.reconfig(cfg)
The rst operation uses rs.conf() (page 988) to set the local variable cfg to the contents of the current replica set conguration, which is a document. The next three operations change the priority (page 464) value in the cfg document for the rst three members congured in the members (page 463) array. The nal operation calls rs.reconfig() (page 989) with the argument of cfg to initialize the new conguration. Note: When updating the replica conguration object, address all members of the set using the index value in the array. The array index begins with 0. Do not confuse this index value with the value of the _id (page 463) eld in each document in the members (page 463) array. The _id (page 463) rarely corresponds to the array index. If a member has priority (page 464) set to 0, it is ineligible to become primary and will not seek election. Hidden members (page 388), delayed members (page 388), and arbiters (page 388) all have priority (page 464) set to 0. All members have a priority (page 464) equal to 1 by default. The value of priority (page 464) can be any oating point (i.e. decimal) number between 0 and 1000. Priorities are only used to determine the preference in election. The priority value is used only in relation to other members. With the exception of members with a priority of 0, the absolute value of the priority (page 464) value is irrelevant. Replica sets will preferentially elect and maintain the primary status of the member with the highest priority (page 464) setting. Warning: Replica set reconguration can force the current primary to step down, leading to an election for primary in the replica set. Elections cause the current primary to close all open client connections. Perform routine replica set reconguration during scheduled maintenance windows. See also: The Replica Reconguration Usage (page 467) example revolves around changing the priorities of the members (page 463) of a replica set.
38.1.7 Resync a Member of a Replica Set

When a secondarys replication process falls so far behind that primary overwrites oplog entries that the secondary has not yet replicated, that secondary cannot catch up and becomes stale. When that occurs, you must completely resynchronize the member by removing its data and performing an initial sync. To do so, use one of the following approaches: Restart the mongod (page 1021) with an empty data directory and let MongoDBs normal initial syncing feature restore the data. This is the more simple option, but may take longer to replace the data. See Automatically Resync a Stale Member (page 429). Restart the machine with a copy of a recent data directory from another member in the replica set. This procedure can replace the data more quickly but requires more manual steps.
428
See Resync by Copying All Datales from Another Member (page 429). Automatically Resync a Stale Member This procedure relies on MongoDBs regular process for initial sync. This will restore the data on the stale member to reect the current state of the set. For an overview of MongoDB initial sync process, see the Syncing (page 411) section. To resync the stale member: 1. Stop the stale members mongod (page 1021) instance. On Linux systems you can use mongod --shutdown (page 1028) Set --dbpath (page 1023) to the members data directory, as in the following:
mongod --dbpath /data/db/ --shutdown
2. Delete all data and sub-directories from the members data directory. By removing the data dbpath (page 1081), MongoDB will perform a complete resync. Consider making a backup rst. 3. Restart the mongod (page 1021) instance on the member. For example:
mongod --dbpath /data/db/ --replSet rsProduction
At this point, the mongod (page 1021) will perform an initial sync. The length of the initial sync may process depends on the size of the database and network connection between members of the replica set. Initial sync operations can impact the other members of the set and create additional trafc to the primary, and can only occur if another member of the set is accessible and up to date. Resync by Copying All Datales from Another Member This approach uses a copy of the data les from an existing member of the replica set, or a backup of the data les to seed the stale member. The copy or backup of the data les must be sufciently recent to allow the new member to catch up with the oplog, otherwise the member would need to perform an initial sync. Note: In most cases you cannot copy data les from a running mongod (page 1021) instance to another, because the data les will change during the le copy operation. Consider the Backup Strategies for MongoDB Systems (page 41) documentation for several methods that you can use to capture a consistent snapshot of a running mongod (page 1021) instance. Important: You must always copy the content of the local database when using data les to resync a member of a replica set. After you have copied the data les from the seed source, start the mongod (page 1021) instance and allow it to apply all operations from the oplog until it reects the current state of the replica set.
38.1.8 Deploy a Geographically Distributed Replica Set

This tutorial outlines the process for deploying a replica set with members in multiple locations. The tutorial addresses three-member sets, four-member sets, and sets with more than four members. For appropriate background, see Replica Set Fundamental Concepts (page 387) and Replica Set Architectures and Deployment Patterns (page 395). For related tutorials, see Deploy a Replica Set (page 419) and Add Members to a Replica Set (page 424).
429
Overview While replica sets provide basic protection against single-instance failure, when all of the members of a replica set reside in a single facility, the replica set is still susceptible to some classes of errors in that facility including power outages, networking distortions, and natural disasters. To protect against these classes of failures, deploy a replica set with one or more members in a geographically distinct facility or data center. Requirements For a three-member replica set you need two instances in a primary facility (hereafter, Site A) and one member in a secondary facility (hereafter, Site B.) Site A should be the same facility or very close to your primary application infrastructure (i.e. application servers, caching layer, users, etc.) For a four-member replica set you need two members in Site A, two members in Site B (or one member in Site B and one member in Site C,) and a single arbiter in Site A. For replica sets with additional members in the secondary facility or with multiple secondary facilities, the requirements are the same as above but with the following notes: Ensure that a majority of the voting members (page 389) are within Site A. This includes secondary-only members (page 388) and arbiters (page 388) For more information on the need to keep the voting majority on one site, see Elections (page 389). If you deploy a replica set with an uneven number of members, deploy an arbiter (page 388) on Site A. The arbiter must be on site A to keep the majority there. For all congurations in this tutorial, deploy each replica set member on a separate system. Although you may deploy more than one replica set member on a single system, doing so reduces the redundancy and capacity of the replica set. Such deployments are typically for testing purposes and beyond the scope of this tutorial. Procedures
Deploy a Distributed Three-Member Replica Set
A geographically distributed three-member deployment has the following features: Each member of the replica set resides on its own machine, and the MongoDB processes all bind to port 27017, which is the standard MongoDB port. Each member of the replica set must be accessible by way of resolvable DNS or hostnames in the following scheme: mongodb0.example.net mongodb1.example.net mongodb2.example.net Congure DNS names appropriately, or set up your systems http://docs.mongodb.org/manual/etc/hosts le to reect this conguration. Ensure that one system (e.g. mongodb2.example.net) resides in Site B. Host all other systems in Site A. Ensure that network trafc can pass between all members in the network securely and efciently. Consider the following: Establish a virtual private network between the systems in Site A and Site B to encrypt all trafc between the sites and remains private. Ensure that your network topology routes all trafc between members within a single site over the local area network.
430
Congure authentication using auth (page 1081) and keyFile (page 1081), so that only servers and process with authentication can connect to the replica set. Congure networking and rewall rules so that only trafc (incoming and outgoing packets) on the default MongoDB port (e.g. 27017) from within your deployment. See also: For more information on security and rewalls, see Security (page 393). Specify run-time conguration on each system in a conguration le (page 1078) stored in http://docs.mongodb.org/manual/etc/mongodb.conf or in a related location. Do not specify run-time conguration through command line options. For each MongoDB instance, use the following conguration, with values set appropriate to your systems:
port = 27017 bind_ip = 10.8.0.10 dbpath = /srv/mongodb/ fork = true replSet = rs0/mongodb0.example.net,mongodb1.example.net,mongodb2.example.net
replSet = rs0
For more documentation on the above run time congurations, as well as additional conguration options, see Conguration File Options (page 1078). To deploy a geographically distributed three-member set: 1. On each system start the mongod (page 1021) process by issuing a command similar to following:
Note: In production deployments you likely want to use and congure a control script to manage this process based on this command. Control scripts are beyond the scope of this document. 2. Open a mongo (page 1036) shell connected to one of the mongod (page 1021) instances:
mongo
3. Use the rs.initiate() (page 989) method on one member to initiate a replica set consisting of the current member and using the default conguration:
rs.initiate()
431
rs.conf()
6. Make sure that you have congured the member located in Site B (i.e. mongodb2.example.net) as a secondary-only member (page 388): (a) Issue the following command to determine the _id (page 463) value for mongodb2.example.net:
rs.conf()
Note: In some situations, the rs.reconfig() (page 989) shell method can force the current primary to step down and causes an election. When the primary steps down, all clients will disconnect. This is the intended behavior. While, this typically takes 10-20 seconds, attempt to make these changes during scheduled maintenance periods. After these commands return you have a geographically distributed three-member replica set. 7. To check the status of your replica set, issue rs.status() (page 990). See also: The documentation of the following shell functions for more information: rs.initiate() (page 989) rs.conf() (page 988) rs.reconfig() (page 989) rs.add() (page 987)
Deploy a Distributed Four-Member Replica Set
A geographically distributed four-member deployment has the following features: Each member of the replica set, except for the arbiter (see below), resides on its own machine, and the MongoDB processes all bind to port 27017, which is the standard MongoDB port. Each member of the replica set must be accessible by way of resolvable DNS or hostnames in the following scheme: mongodb0.example.net mongodb1.example.net mongodb2.example.net
432
mongodb3.example.net Congure DNS names appropriately, or set up your systems http://docs.mongodb.org/manual/etc/host le to reect this conguration. Ensure that one system (e.g. mongodb2.example.net) resides in Site B. Host all other systems in Site A. One host (e.g. mongodb3.example.net) will be an arbiter and can run on a system that is also used for an application server or some other shared purpose. There are three possible architectures for this replica set: Two members in Site A, two secondary-only members (page 388) in Site B, and an arbiter in Site A. Three members in Site A and one secondary-only member in Site B. Two members in Site A, one secondary-only member in Site B, one secondary-only member in Site C, and an arbiter in site A. In most cases the rst architecture is preferable because it is the least complex. Ensure that network trafc can pass between all members in the network securely and efciently. Consider the following: Establish a virtual private network between the systems in Site A and Site B (and Site C if it exists) to encrypt all trafc between the sites and remains private. Ensure that your network topology routes all trafc between members within a single site over the local area network. Congure authentication using auth (page 1081) and keyFile (page 1081), so that only servers and process with authentication can connect to the replica set. Congure networking and rewall rules so that only trafc (incoming and outgoing packets) on the default MongoDB port (e.g. 27017) from within your deployment. See also: For more information on security and rewalls, see Security (page 393). Specify run-time conguration on each system in a conguration le (page 1078) stored in http://docs.mongodb.org/manual/etc/mongodb.conf or in a related location. Do not specify run-time conguration through command line options. For each MongoDB instance, use the following conguration, with values set appropriate to your systems:
port = 27017 bind_ip = 10.8.0.10 dbpath = /srv/mongodb/ fork = true
replSet = rs0/mongodb0.example.net,mongodb1.example.net,mongodb2.example.net,mongodb3.example.ne
433
replSet = rs0
For more documentation on the above run time congurations, as well as additional conguration options, see doc:/reference/conguration-options. To deploy a geographically distributed four-member set: 1. On each system start the mongod (page 1021) process by issuing a command similar to following:
mongo
rs.initiate()

rs.conf()
rs.add("mongodb1.example.net") rs.add("mongodb2.example.net") rs.add("mongodb3.example.net")
6. In the same shell session, issue the following command to add the arbiter (e.g. mongodb4.example.net):
rs.addArb("mongodb4.example.net")
7. Make sure that you have congured each member located in Site B (e.g. mongodb3.example.net) as a secondary-only member (page 388): (a) Issue the following command to determine the _id (page 463) value for the member:
rs.conf()
Note: In some situations, the rs.reconfig() (page 989) shell method can force the current primary to step down and causes an election. When the primary steps down, all clients will disconnect. This is the intended behavior. While, this typically takes 10-20 seconds, attempt to make these changes during scheduled maintenance periods.
434
After these commands return you have a geographically distributed four-member replica set. 8. To check the status of your replica set, issue rs.status() (page 990). See also: The documentation of the following shell functions for more information: rs.initiate() (page 989) rs.conf() (page 988) rs.reconfig() (page 989) rs.add() (page 987)
Deploy a Distributed Set with More than Four Members
The procedure for deploying a geographically distributed set with more than four members is similar to the above procedures, with the following differences: Never deploy more than seven voting members. Use the procedure for a four-member set if you have an even number of members (see Deploy a Distributed Four-Member Replica Set (page 432)). Ensure that Site A always has a majority of the members by deploying the arbiter within Site A. For six member sets, deploy at least three voting members in addition to the arbiter in Site A, the remaining members in alternate sites. Use the procedure for a three-member set if you have an odd number of members (see Deploy a Distributed Three-Member Replica Set (page 430)). Ensure that Site A always has a majority of the members of the set. For example, if a set has ve members, deploy three members within the primary facility and two members in other facilities. If you have a majority of the members of the set outside of Site A and the network partitions to prevent communication between sites, the current primary in Site A will step down, even if none of the members outside of Site A are eligible to become primary.
38.1.9 Change the Size of the Oplog

The oplog exists internally as a capped collection, so you cannot modify its size in the course of normal operations. In most cases the default oplog size (page 392) is an acceptable size; however, in some situations you may need a larger or smaller oplog. For example, you might need to change the oplog size if your applications perform large numbers of multi-updates or deletes in short periods of time. This tutorial describes how to resize the oplog. For a detailed explanation of oplog sizing, see the Oplog (page 392) topic in the Replica Set Fundamental Concepts (page 387) document. For details on the how oplog size affects delayed members and affects replication lag, see the Delayed Members (page 388) topic and Check the Replication Lag (page 444). Overview The following is an overview of the procedure for changing the size of the oplog: 1. Shut down one of the secondary members of your replica set and then restart it on a different port and in standalone mode. 2. Create a backup of the old (current) oplog. This is optional. 3. Save the last entry from the old oplog.
435
4. Drop the old oplog. 5. Create a new oplog of a different size. 6. Insert the previously saved last entry from the old oplog into the new oplog. 7. Restart the server as a member of the replica set on its usual port. 8. Apply this procedure to all secondaries that could become primary. 9. Step down the current primary with rs.stepDown() (page 991), and repeat oplog resizing process above for the former primary (e.g. steps 2 through 7). Procedure The examples in this procedure use the following conguration: The active replica set is rs0. The replica set is running on port 27017. The replica set is running with a data directory http://docs.mongodb.org/manual/srv/mongodb. (page 1081) of
To change the size of the oplog for a replica set, use the following procedure for every member of the set that may become primary. 1. Shut down the mongod (page 1021) instance and restart it in standalone mode running on a different port. Note: Shutting down the primary member of the set will trigger a failover situation and another member in the replica set will become primary. In most cases, it is least disruptive to modify the oplogs of all the secondaries before modifying the primary. To shut down the current primary instance, use a command that resembles the following:
mongod --dbpath /srv/mongodb --shutdown
To restart the instance on a different port and in standalone mode (i.e. without replSet (page 1088) or --replSet (page 1028)), use a command that resembles the following:
mongod --port 37017 --dbpath /srv/mongodb
2. Backup the existing oplog on the standalone instance. Use the following sequence of commands:
mongodump --db local --collection oplog.rs --port 37017
Note: You can restore the backup using the mongorestore (page 1048) utility. Connect to the instance using the mongo (page 1036) shell:
mongo --port 37017
3. Save the last entry from the old (current) oplog. (a) In the mongo (page 1036) shell, enter the following command to use the local database to interact with the oplog:
use local
(b) Use the db.collection.save() (page 946) operation to save the last entry in the oplog to a temporary collection:
436
db.temp.save( db.oplog.rs.find( { }, { ts: 1, h: 1 } ).sort( {$natural : -1} ).limit(1).next
You can see this oplog entry in the temp collection by issuing the following command:
db.temp.find()
4. Drop the old oplog.rs collection in the local database. Use the following command:
db.oplog.rs.drop()
This will return true on the shell. 5. Use the create (page 863) command to create a new oplog of a different size. Specify the size argument in bytes. A value of 2147483648 will create a new oplog thats 2 gigabytes:
db.runCommand( { create : "oplog.rs", capped : true, size : 2147483648 } )
Upon success, this command returns the following status:

{ "ok" : 1 }
6. Insert the previously saved last entry from the old oplog into the new oplog:
db.oplog.rs.save( db.temp.findOne() )
To conrm the entry is in the new oplog, issue the following command:
db.oplog.rs.find()
7. Restart the server as a member of the replica set on its usual port:
mongod --dbpath /srv/mongodb --shutdown mongod --replSet rs0 --dbpath /srv/mongodb
The replica member will recover and catch up and then will be eligible for election to primary. To step down the temporary primary that took over when you initially shut down the server, use the rs.stepDown() (page 991) method. This will force an election for primary. If the servers priority (page 390) is higher than all other members in the set and if it has successfully caught up, then it will likely become primary. 8. Repeat this procedure for all other members of the replica set that are or could become primary.
38.1.10 Force a Member to Become Primary

Synopsis You can force a replica set member to become primary by giving it a higher priority (page 464) value than any other member in the set. Optionally, you also can force a member never to become primary by setting its priority (page 464) value to 0, which means the member can never seek election (page 389) as primary. For more information, see Secondary-Only Members (page 388). Procedures
Force a Member to be Primary by Setting its Priority High
Changed in version 2.0.
437
For more information on priorities, see Member Priority (page 390). This procedure assumes your current primary is m1.example.net and that youd like to instead make m3.example.net primary. The procedure also assumes you have a three-member replica set with the conguration below. For more information on congurations, see Replica Set Conguration Use (page 467). This procedure assumes this conguration:
{ "_id" : "rs", "version" : 7, "members" : [ { "_id" : 0, "host" : "m1.example.net:27017" }, { "_id" : 1, "host" : "m2.example.net:27017" }, { "_id" : 2, "host" : "m3.example.net:27017" } ] }
1. In the mongo (page 1036) shell, use the following sequence of operations to make m3.example.net the primary:
cfg = rs.conf() cfg.members[0].priority = 0.5 cfg.members[1].priority = 0.5 cfg.members[2].priority = 1 rs.reconfig(cfg)
This sets m3.example.net to have a higher local.system.replset.members[n].priority (page 464) value than the other mongod (page 1021) instances. The following sequence of events occur: m3.example.net and m2.example.net sync with m1.example.net (typically within 10 seconds). m1.example.net sees that it no longer has highest priority and, in most cases, steps down. m1.example.net does not step down if m3.example.nets sync is far behind. In that case, m1.example.net waits until m3.example.net is within 10 seconds of its optime and then steps down. This minimizes the amount of time with no primary following failover. The step down forces on election in which m3.example.net becomes primary based on its priority (page 464) setting. 2. Optionally, if m3.example.net is more than 10 seconds behind m1.example.nets optime, and if you dont need to have a primary designated within 10 seconds, you can force m1.example.net to step down by running:
db.adminCommand({replSetStepDown:1000000, force:1})
This prevents m1.example.net from being primary for 1,000,000 seconds, even if there is no other member that can become primary. When m3.example.net catches up with m1.example.net it will become primary.
438
If you later want to make m1.example.net primary again while it waits for m3.example.net to catch up, issue the following command to make m1.example.net seek election again:
rs.freeze()
The rs.freeze() (page 988) provides a wrapper around the replSetFreeze (page 840) database command.
Force a Member to be Primary Using Database Commands
Changed in version 1.8. Consider a replica set with the following members: mdb0.example.net - the current primary. mdb1.example.net - a secondary. mdb2.example.net - a secondary . To force a member to become primary use the following procedure: 1. In a mongo (page 1036) shell, run rs.status() (page 990) to ensure your replica set is running as expected. 2. In a mongo (page 1036) shell connected to the mongod (page 1021) instance running on mdb2.example.net, freeze mdb2.example.net so that it does not attempt to become primary for 120 seconds.
rs.freeze(120)
3. In a mongo (page 1036) shell connected the mongod (page 1021) running on mdb0.example.net, step down this instance that the mongod (page 1021) is not eligible to become primary for 120 seconds:
rs.stepDown(120)
mdb1.example.net becomes primary. Note: During the transition, there is a short window where the set does not have a primary. For more information, consider the rs.freeze() (page 988) and rs.stepDown() (page 991) methods that wrap the replSetFreeze (page 840) and replSetStepDown (page 844) commands.
38.1.11 Manage Chained Replication

Starting in version 2.0, MongoDB supports chained replication. A chained replication occurs when a secondary member replicates from another secondary member instead of from the primary. This might be the case, for example, if a secondary selects its replication target based on ping time and if the closest member is another secondary. Chained replication can reduce load on the primary. But chained replication can also result in increased replication lag, depending on the topology of the network. New in version 2.2.2. You can use the chainingAllowed (page 465) setting in Replica Set Conguration (page 463) to disable chained replication for situations where chained replication is causing lag. MongoDB enables chained replication by default. This procedure describes how to disable it and how to re-enable it. Note: If chained replication is disabled, you still can use replSetSyncFrom (page 845) to specify that a secondary
439
replicates from another secondary. But that conguration will last only until the secondary recalculates which member to sync from.
Disable Chained Replication To disable chained replication, set the chainingAllowed (page 465) eld in Replica Set Conguration (page 463) to false. You can use the following sequence of commands to set chainingAllowed (page 465) to false: 1. Copy the conguration settings into the cfg object:
cfg = rs.config()
2. Take note of whether the current conguration settings contain the settings sub-document. If they do, skip this step. Warning: document. To avoid data loss, skip this step if the conguration settings contain the settings sub-
If the current conguration settings do not contain the settings sub-document, create the sub-document by issuing the following command:
cfg.settings = { }
3. Issue the following sequence of commands to set chainingAllowed (page 465) to false:
cfg.settings.chainingAllowed = false rs.reconfig(cfg)
Re-enable Chained Replication To re-enable chained replication, set chainingAllowed (page 465) to true. You can use the following sequence of commands:
cfg = rs.config() cfg.settings.chainingAllowed = true rs.reconfig(cfg)
38.1.12 Change Hostnames in a Replica Set

Synopsis For most replica sets the hostnames 3 in the host (page 463) eld never change. However, in some cases you must migrate some or all host names in a replica set as organizational needs change. This document presents two possible procedures for changing the hostnames in the host (page 463) eld. Depending on your environments availability requirements, you may:
3
Always use resolvable hostnames for the value of the host (page 463) eld in the replica set conguration to avoid confusion and complexity.
440
1. Make the conguration change without disrupting the availability of the replica set. While this ensures that your application will always be able to read and write data to the replica set, this procedure can take a long time and may incur downtime at the application layer. 4 For this procedure, see Changing Hostnames while Maintaining the Replica Sets Availability (page 442). 2. Stop all members of the replica set at once running on the old hostnames or interfaces, make the conguration changes, and then start the members at the new hostnames or interfaces. While the set will be totally unavailable during the operation, the total maintenance window is often shorter. For this procedure, see Changing All Hostnames in Replica Set at Once (page 443). See also: Replica Set Conguration (page 463) Replica Set Reconguration Process (page 467) rs.conf() (page 988) and rs.reconfig() (page 989) And the following tutorials: Deploy a Replica Set (page 419) Add Members to a Replica Set (page 424) Procedures Given a replica set with three members: database0.example.com:27017 (the primary) database1.example.com:27017 database2.example.com:27017 And with the following rs.conf() (page 988) output:
{ "_id" : "rs", "version" : 3, "members" : [ { "_id" : 0, "host" : "database0.example.com:27017" }, { "_id" : 1, "host" : "database1.example.com:27017" }, { "_id" : 2, "host" : "database2.example.com:27017" } ] }
The following procedures change the members hostnames as follows: mongodb0.example.net:27017 (the primary)
4 You will have to congure your applications so that they can connect to the replica set at both the old and new locations. This often requires a restart and reconguration at the application layer, which may affect the availability of your applications. This re-conguration is beyond the scope of this document and makes the second option (page 443) preferable when you must change the hostnames of all members of the replica set at once.
441
mongodb1.example.net:27017 mongodb2.example.net:27017 Use the most appropriate procedure for your deployment.
Changing Hostnames while Maintaining the Replica Sets Availability
This procedure uses the above assumptions (page 441). 1. For each secondary in the replica set, perform the following sequence of operations: (a) Stop the secondary. (b) Restart the secondary at the new location. (c) Open a mongo (page 1036) shell connected to the replica sets primary. In our example, the primary runs on port 27017 so you would issue the following command:
mongo --port 27017
(d) Run the following recongure option, for the host (page 463) value where n is 1:
cfg = rs.conf() cfg.members[1].host = "mongodb1.example.net:27017" rs.reconfig(cfg)
See Replica Set Conguration (page 463) for more information. (e) Make sure your client applications are able to access the set at the new location and that the secondary has a chance to catch up with the other members of the set. Repeat the above steps for each non-primary member of the set. 2. Open a mongo (page 1036) shell connected to the primary and step down the primary using replSetStepDown (page 844). In the mongo (page 1036) shell, use the rs.stepDown() (page 991) wrapper, as follows:
rs.stepDown()
3. When the step down succeeds, shut down the primary. 4. To make the nal conguration change, connect to the new primary in the mongo (page 1036) shell and recongure the host (page 463) value where n is 0:
cfg = rs.conf() cfg.members[0].host = "mongodb0.example.net:27017" rs.reconfig(cfg)
5. Start the original primary. 6. Open a mongo (page 1036) shell connected to the primary. 7. To conrm the new conguration, call rs.conf() (page 988) in the mongo (page 1036) shell. Your output should resemble:
442
{ "_id" : "rs", "version" : 4, "members" : [ { "_id" : 0, "host" : "mongodb0.example.net:27017" }, { "_id" : 1, "host" : "mongodb1.example.net:27017" }, { "_id" : 2, "host" : "mongodb2.example.net:27017" } ] }
Changing All Hostnames in Replica Set at Once
This procedure uses the above assumptions (page 441). 1. Stop all members in the replica set. 2. Restart each member on a different port and without using the --replSet (page 1028) run-time option. Changing the port number during maintenance prevents clients from connecting to this host while you perform maintenance. Use the members usual --dbpath (page 1023), which in this example is http://docs.mongodb.org/manual/data/db1. Use a command that resembles the following:
mongod --dbpath /data/db1/ --port 37017
3. For each member of the replica set, perform the following sequence of operations: (a) Open a mongo (page 1036) shell connected to the mongod (page 1021) running on the new, temporary port. For example, for a member running on a temporary port of 37017, you would issue this command:
mongo --port 37017
(b) Edit the replica set conguration manually. The replica set conguration is the only document in the system.replset collection in the local database. Edit the replica set conguration with the new hostnames and correct ports for all the members of the replica set. Consider the following sequence of commands to change the hostnames in a three-member set:
use local cfg = db.system.replset.findOne( { "_id": "rs" } ) cfg.members[0].host = "mongodb0.example.net:27017" cfg.members[1].host = "mongodb1.example.net:27017" cfg.members[2].host = "mongodb2.example.net:27017" db.system.replset.update( { "_id": "rs" } , cfg )
(c) Stop the mongod (page 1021) process on the member.
443
4. After re-conguring all members of the set, start each mongod (page 1021) instance in the normal way: use the usual port number and use the --replSet (page 1028) option. For example:
mongod --dbpath /data/db1/ --port 27017 --replSet rs
5. Connect to one of the mongod (page 1021) instances using the mongo (page 1036) shell. For example:
mongo --port 27017
6. To conrm the new conguration, call rs.conf() (page 988) in the mongo (page 1036) shell. Your output should resemble:
{ "_id" : "rs", "version" : 4, "members" : [ { "_id" : 0, "host" : "mongodb0.example.net:27017" }, { "_id" : 1, "host" : "mongodb1.example.net:27017" }, { "_id" : 2, "host" : "mongodb2.example.net:27017" } ] }
38.1.13 Troubleshoot Replica Sets

This section describes common strategies for troubleshooting replica sets. Check Replica Set Status To display the current state of the replica set and current state of each member, run the rs.status() (page 990) method in a mongo (page 1036) shell connected to the replica sets primary. For descriptions of the information displayed by rs.status() (page 990), see replSetGetStatus (page 840). Note: The rs.status() (page 990) method is a wrapper that runs the replSetGetStatus (page 840) database command.
Check the Replication Lag Replication lag is a delay between an operation on the primary and the application of that operation from the oplog to the secondary. Replication lag can be a signicant issue and can seriously affect MongoDB replica set deployments. Excessive replication lag makes lagged members ineligible to quickly become primary and increases the possibility that distributed read operations will be inconsistent. To check the current length of replication lag:
444
In a mongo (page 1036) shell connected to the primary, call the db.printSlaveReplicationInfo() (page 983) method. The returned document displays the syncedTo value for each member, which shows you when each member last read from the oplog, as shown in the following example:
source: m1.example.net:30001 syncedTo: Tue Oct 02 2012 11:33:40 GMT-0400 (EDT) = 7475 secs ago (2.08hrs) source: m2.example.net:30002 syncedTo: Tue Oct 02 2012 11:33:40 GMT-0400 (EDT) = 7475 secs ago (2.08hrs)
Note: The rs.status() (page 990) method is a wrapper around the replSetGetStatus (page 840) database command. Monitor the rate of replication by watching the oplog time in the replica graph in the MongoDB Monitoring Service. For more information see the documentation for MMS. Possible causes of replication lag include: Network Latency Check the network routes between the members of your set to ensure that there is no packet loss or network routing issue. Use tools including ping to test latency between set members and traceroute to expose the routing of packets network endpoints. Disk Throughput If the le system and disk device on the secondary is unable to ush data to disk as quickly as the primary, then the secondary will have difculty keeping state. Disk-related issues are incredibly prevalent on multitenant systems, including vitalized instances, and can be transient if the system accesses disk devices over an IP network (as is the case with Amazons EBS system.) Use system-level tools to assess disk status, including iostat or vmstat. Concurrency In some cases, long-running operations on the primary can block replication on secondaries. For best results, congure write concern (page 398) to require conrmation of replication to secondaries, as described in replica set write concern (page 399). This prevents write operations from returning if replication cannot keep up with the write load. Use the database proler to see if there are slow queries or long-running operations that correspond to the incidences of lag. Appropriate Write Concern If you are performing a large data ingestion or bulk load operation that requires a large number of writes to the primary, particularly with unacknowledged write concern (page 399), the secondaries will not be able to read the oplog fast enough to keep up with changes. To prevent this, require write acknowledgment or journaled write concern (page 180) after every 100, 1,000, or an another interval to provide an opportunity for secondaries to catch up with the primary. For more information see: Replica Acknowledge Write Concern (page 399) Replica Set Write Concern (page 183)
445
Oplog (page 392) Test Connections Between all Members All members of a replica set must be able to connect to every other member of the set to support replication. Always verify connections in both directions. Networking topologies and rewall congurations prevent normal and required connectivity, which can block replication. Consider the following example of a bidirectional test of networking: Example Given a replica set with three members running on three separate hosts: m1.example.net m2.example.net m3.example.net 1. Test the connection from m1.example.net to the other hosts with the following operation set m1.example.net:
mongo --host m2.example.net --port 27017 mongo --host m3.example.net --port 27017
2. Test the connection from m2.example.net to the other two hosts with the following operation set from m2.example.net, as in:
You have now tested the connection between m2.example.net and m1.example.net in both directions. 3. Test the connection from m3.example.net to the other two hosts with the following operation set from the m3.example.net host, as in:
If any connection, in any direction fails, check your networking and rewall conguration and recongure your environment to allow these connections.
Check the Size of the Oplog A larger oplog can give a replica set a greater tolerance for lag, and make the set more resilient. To check the size of the oplog for a given replica set member, connect to the member in a mongo (page 1036) shell and run the db.printReplicationInfo() (page 983) method. The output displays the size of the oplog and the date ranges of the operations contained in the oplog. In the following example, the oplog is about 10MB and is able to t about 26 hours (94400 seconds) of operations:
configured oplog size: 10.10546875MB log length start to end: 94400 (26.22hrs) oplog first event time: Mon Mar 19 2012 13:50:38 GMT-0400 (EDT)
446
oplog last event time: now:
Wed Oct 03 2012 14:59:10 GMT-0400 (EDT) Wed Oct 03 2012 15:00:21 GMT-0400 (EDT)
The oplog should be long enough to hold all transactions for the longest downtime you expect on a secondary. At a minimum, an oplog should be able to hold minimum 24 hours of operations; however, many users prefer to have 72 hours or even a weeks work of operations. For more information on how oplog size affects operations, see: Oplog (page 392), Delayed Members (page 388), and Check the Replication Lag (page 444). Note: You normally want the oplog to be the same size on all members. If you resize the oplog, resize it on all members. To change oplog size, see the Change the Size of the Oplog (page 435) tutorial. Oplog Entry Timestamp Error Consider the following error in mongod (page 1021) output and logs:
replSet error fatal couldnt query the local local.oplog.rs collection. <timestamp> [rsStart] bad replSet oplog entry? Terminating mongod after 30
Often, an incorrectly typed value in the ts eld in the last oplog entry causes this error. The correct data type is Timestamp. Check the type of the ts value using the following two queries against the oplog collection:
db = db.getSiblingDB("local") db.oplog.rs.find().sort({$natural:-1}).limit(1) db.oplog.rs.find({ts:{$type:17}}).sort({$natural:-1}).limit(1)
The rst query returns the last document in the oplog, while the second returns the last document in the oplog where the ts value is a Timestamp. The $type (page 773) operator allows you to select BSON type 17, is the Timestamp data type. If the queries dont return the same document, then the last document in the oplog has the wrong data type in the ts eld. Example If the rst query returns this as the last oplog entry:
{ "ts" : {t: 1347982456000, i: 1}, "h" : NumberLong("8191276672478122996"), "op" : "n", "ns" : "", "o" : { "msg" : "Reconfig set", "version" : 4 } }
And the second query returns this as the last entry where ts has the Timestamp type:
{ "ts" : Timestamp(1347982454000, 1), "h" : NumberLong("6188469075153256465"), "op" : "n", "ns" : "", "o" : { "msg" : "Reconfig set", "version" : 3 } }
447
Then the value for the ts eld in the last oplog entry is of the wrong data type. To set the proper type for this value and resolve this issue, use an update operation that resembles the following:
db.oplog.rs.update( { ts: { t:1347982456000, i:1 } }, { $set: { ts: new Timestamp(1347982456000, 1)}})
Modify the timestamp values as needed based on your oplog entry. This operation may take some period to complete because the update must scan and pull the entire oplog into memory. Duplicate Key Error on local.slaves The duplicate key on local.slaves error, occurs when a secondary or slave changes its hostname and the primary or master tries to update its local.slaves collection with the new name. The update fails because it contains the same _id value as the document containing the previous hostname. The error itself will resemble the following.
exception 11000 E11000 duplicate key error index: local.slaves.$_id_ dup key: { : ObjectId(<object
This is a benign error and does not affect replication operations on the secondary or slave. To prevent the error from appearing, drop the local.slaves collection from the primary or master, with the following sequence of operations in the mongo (page 1036) shell:
use local db.slaves.drop()
The next time a secondary or slave polls the primary or master, the primary or master recreates the local.slaves collection.
38.1.14 Add an Arbiter to Replica Set

Arbiters are special mongod (page 1021) instances that do not hold a copy of the data and thus cannot become primary. Arbiters exist solely to participate in elections (page 389). Because arbiters do not hold a copies of collection data, they have minimal resource requirements and do not require dedicated hardware. Note: Because of their minimal system requirements, you may safely deploy an arbiter on a system with another workload, such as an application server or monitoring member. Warning: Do not run arbiter processes on a system that is an active primary or secondary of its replica set.
Add an Arbiter Note: To prevent tied elections, do not add an arbiter to a set if the set already has an odd number of voting members. 1. Create a data directory for the arbiter. The mongod (page 1021) uses this directory for conguration information. It will not hold database collection data. The following example creates the http://docs.mongodb.org/manual/data/arb data directory:
mkdir /data/arb
2. Start the arbiter, making sure to specify the replica set name and the data directory. Consider the following example:
448
mongod --port 30000 --dbpath /data/arb --replSet rs
3. In a mongo (page 1036) shell connected to the primary, add the arbiter to the replica set by issuing the rs.addArb() (page 988) method, which uses the following syntax:
rs.addArb("<hostname><:port>")
For example, if the arbiter runs on m1.example.net:30000, you would issue this command:
rs.addArb("m1.example.net:30000")
38.1.15 Convert a Secondary to an Arbiter

If you have a secondary in a replica set that no longer needs to hold a copy of the data but you want to retain it in the set to ensure that the replica set will be able to elect a primary (page 389), you can convert the secondary into an arbiter (page 388). This document provides two equivalent procedures for this process. Synopsis Both of the following procedures are operationally equivalent. Choose whichever procedure you are most comfortable with: 1. You may operate the arbiter on the same port as the former secondary. In this procedure, you must shut down the secondary and remove its data before restarting and reconguring it as an arbiter. For this procedure, see Convert a Secondary to an Arbiter and Reuse the Port Number (page 449). 2. Run the arbiter on a new port. In this procedure, you can recongure the server as an arbiter before shutting down the instance running as a secondary. For this procedure, see Convert a Secondary to an Arbiter Running on a New Port Number (page 450). See also: Arbiters (page 388) rs.addArb() (page 988) Replica Set Administration (page 419) Procedures
Convert a Secondary to an Arbiter and Reuse the Port Number
1. If your application is connecting directly to the secondary, modify the application so that MongoDB queries dont reach the secondary. 2. Shut down the secondary. 3. Remove the secondary from the replica set by calling the rs.remove() (page 990) method. Perform this operation while connected to the current primary in the mongo (page 1036) shell:
rs.remove("<hostname><:port>")
4. Verify that the replica set no longer includes the secondary by calling the rs.conf() (page 988) method in the mongo (page 1036) shell:
449
rs.conf()
5. Move the secondarys data directory to an archive folder. For example:

mv /data/db /data/db-old
Optional You may remove the data instead. 6. Create a new, empty data directory to point to when restarting the mongod (page 1021) instance. You can reuse the previous name. For example:
mkdir /data/db
7. Restart the mongod (page 1021) instance for the secondary, specifying the port number, the empty data directory, and the replica set. You can use the same port number you used before. Issue a command similar to the following:
mongod --port 27021 --dbpath /data/db --replSet rs
8. In the mongo (page 1036) shell convert the secondary to an arbiter using the rs.addArb() (page 988) method:
9. Verify the arbiter belongs to the replica set by calling the rs.conf() (page 988) method in the mongo (page 1036) shell.
rs.conf()
The arbiter member should include the following:

"arbiterOnly" : true
Convert a Secondary to an Arbiter Running on a New Port Number
1. If your application is connecting directly to the secondary or has a connection string referencing the secondary, modify the application so that MongoDB queries dont reach the secondary. 2. Create a new, empty data directory to be used with the new port number. For example:
mkdir /data/db-temp
3. Start a new mongod (page 1021) instance on the new port number, specifying the new data directory and the existing replica set. Issue a command similar to the following:
mongod --port 27021 --dbpath /data/db-temp --replSet rs
4. In the mongo (page 1036) shell connected to the current primary, convert the new mongod (page 1021) instance to an arbiter using the rs.addArb() (page 988) method:
5. Verify the arbiter has been added to the replica set by calling the rs.conf() (page 988) method in the mongo (page 1036) shell.
450
rs.conf()
The arbiter member should include the following:

"arbiterOnly" : true
6. Shut down the secondary. 7. Remove the secondary from the replica set by calling the rs.remove() (page 990) method in the mongo (page 1036) shell:
rs.remove("<hostname><:port>")
8. Verify that the replica set no longer includes the old secondary by calling the rs.conf() (page 988) method in the mongo (page 1036) shell:
rs.conf()
9. Move the secondarys data directory to an archive folder. For example:

Optional You may remove the data instead.
38.1.16 Congure a Delayed Replica Set Member

To congure a replica set member with a one hour delay, use the following sequence of operations in the mongo (page 1036) shell:
cfg = rs.conf() cfg.members[0].priority = 0 cfg.members[0].slaveDelay = 3600 rs.reconfig(cfg)
Warning: The rs.reconfig() (page 989) shell method can force the current primary to step down, which causes an election (page 389). When the primary steps down, the mongod (page 1021) closes all client connections. While this typically takes 10-20 seconds, attempt to make these changes during scheduled maintenance periods. To successfully recongure a replica set, a majority of the members must be accessible. After the replica set recongures, the rst member of the set in the members (page 463) array will have a priority of 0 and cannot become primary. The slaveDelay value delays both replication and the members oplog by 3600 seconds (1 hour). Setting slaveDelay (page 465) to a non-zero value also sets hidden (page 464) to true for this replica set so that it does not receive application queries in normal operations. See also: The Elections (page 389) section in the Replica Set Fundamental Concepts (page 387) document, and the Election Internals (page 409) section in the Replica Set Internals and Behaviors (page 408) document. Warning: The length of the secondary slaveDelay (page 465) must t within the window of the oplog. If the oplog is shorter than the slaveDelay (page 465) window, the delayed member cannot successfully replicate operations. See also:
451
slaveDelay (page 465), Replica Set Reconguration (page 467), Oplog (page 392), and the Change the Size of the Oplog (page 435) tutorial.
38.1.17 Congure a Replica Set Member as Hidden

Hidden members are part of a replica set but cannot become primary and are invisible to client applications. However, hidden members do vote in elections (page 389). Hidden members are ideal for instances that will have signicantly different usage patterns than the other members and require separation from normal trafc. Typically, hidden members provide reporting, dedicated backups, and dedicated read-only testing and integration support. Hidden members have priority (page 464) set 0 and have hidden (page 464) set to true. To congure a hidden member, on the primary, use the following sequence of operations in the mongo (page 1036) shell:
cfg = rs.conf() cfg.members[0].priority = 0 cfg.members[0].hidden = true rs.reconfig(cfg)
Warning: The rs.reconfig() (page 989) shell method can force the current primary to step down, which causes an election (page 389). When the primary steps down, the mongod (page 1021) closes all client connections. While this typically takes 10-20 seconds, attempt to make these changes during scheduled maintenance periods. To successfully recongure a replica set, a majority of the members must be accessible. After re-conguring the set, the rst member of the set in the members (page 463) array will have a priority of 0 so that it cannot become primary. The other members in the set will not advertise the hidden member in the isMaster (page 838) or db.isMaster() (page 981) output. See also: The Elections (page 389) section in the Replica Set Fundamental Concepts (page 387) document, and the Election Internals (page 409) section in the Replica Set Internals and Behaviors (page 408) document. Changed in version 2.0: For sharded clusters running with replica sets before 2.0, if you recongured a member as hidden, you had to restart mongos (page 1032) to prevent queries from reaching the hidden member. See also: Read Preference (page 402) and Replica Set Reconguration (page 467).
38.1.18 Congure a Non-Voting Replica Set Member

You may choose to change the number of votes that each member has in elections (page 389) for primary. In general, all members should have only 1 vote to prevent intermittent ties, deadlock, or the wrong members from becoming primary. Use replica set priorities (page 390) to control which members are more likely to become primary. To disable a members ability to vote in elections, use the following command sequence in the mongo (page 1036) shell.
cfg = rs.conf() cfg.members[3].votes = 0 cfg.members[4].votes = 0 cfg.members[5].votes = 0 rs.reconfig(cfg)
452
This sequence gives 0 votes to the fourth, fth, and sixth members of the set according to the order of the members (page 463) array in the output of rs.conf() (page 988). This setting allows the set to elect these members as primary but does not allow them to vote in elections. If you have three non-voting members, you can add three additional voting members to your set. Place voting members so that your designated primary or primaries can reach a majority of votes in the event of a network partition. Warning: The rs.reconfig() (page 989) shell method can force the current primary to step down, which causes an election (page 389). When the primary steps down, the mongod (page 1021) closes all client connections. While this typically takes 10-20 seconds, attempt to make these changes during scheduled maintenance periods. To successfully recongure a replica set, a majority of the members must be accessible.
Note: In general and when possible, all members should have only 1 vote. This prevents intermittent ties, deadlocks, or the wrong members from becoming primary. Use Replica Set Priorities (page 390) to control which members are more likely to become primary. See also: votes (page 465) and Replica Set Reconguration (page 467). See also: The Elections (page 389) section in the Replica Set Fundamental Concepts (page 387) document, and the Election Internals (page 409) section in the Replica Set Internals and Behaviors (page 408) document.
38.1.19 Prevent Replica Set Member from Becoming Primary

The secondary-only conguration prevents a secondary member in a replica set from ever becoming a primary in a failover. You can set secondary-only mode for any member of the set except the current primary. For example, you may want to congure all members of a replica sets located outside of the main data centers as secondary-only to prevent these members from ever becoming primary. To congure a member as secondary-only, set its priority (page 464) value to 0. Any member with a priority (page 464) equal to 0 will never seek election (page 389) and cannot become primary in any situation. For more information on priority levels, see Member Priority (page 390). Note: When updating the replica conguration object, address all members of the set using the index value in the array. The array index begins with 0. Do not confuse this index value with the value of the _id (page 463) eld in each document in the members (page 463) array. The _id (page 463) rarely corresponds to the array index. As an example of modifying member priorities, assume a four-member replica set. Use the following sequence of operations in the mongo (page 1036) shell to modify member priorities:
cfg = rs.conf() cfg.members[0].priority cfg.members[1].priority cfg.members[2].priority cfg.members[3].priority rs.reconfig(cfg) = = = = 2 1 0.5 0
This recongures the set, with the following priority settings: Member 0 to a priority of 2 so that it becomes primary, under most circumstances.
453
Member 1 to a priority of 1, which is the default value. Member 1 becomes primary if no member with a higher priority is eligible. Member 2 to a priority of 0.5, which makes it less likely to become primary than other members but doesnt prohibit the possibility. Member 3 to a priority of 0. Member 3 cannot become the primary member under any circumstances. Warning: The rs.reconfig() (page 989) shell method can force the current primary to step down, which causes an election (page 389). When the primary steps down, the mongod (page 1021) closes all client connections. While this typically takes 10-20 seconds, attempt to make these changes during scheduled maintenance periods. To successfully recongure a replica set, a majority of the members must be accessible.
Note: If your replica set has an even number of members, add an arbiter (page 448) to ensure that members can quickly obtain a majority of votes in an election for primary. MongoDB does not permit the current primary to have a priority (page 464) of 0. If you want to prevent the current primary from becoming primary, rst use rs.stepDown() (page 991) to step down the current primary, and then recongure the replica set (page 467) with rs.conf() (page 988) and rs.reconfig() (page 989). See also: priority (page 464) and Replica Set Reconguration (page 467).
38.1.20 Congure a Secondarys Sync Target

To override the default sync target selection logic, you may manually congure a secondary members sync target for pulling oplog entries temporarily. The following operations provide access to this functionality: replSetSyncFrom (page 845) command, or rs.syncFrom() (page 991) helper in the mongo (page 1036) shell Only modify the default sync logic as needed, and always exercise caution. rs.syncFrom() (page 991) will not affect an in-progress initial sync operation. To affect the sync target for the initial sync, run rs.syncFrom() (page 991) operation before initial sync. If you run rs.syncFrom() (page 991) during initial sync, MongoDB produces no error messages, but the sync target will not change until after the initial sync operation. Note: replSetSyncFrom (page 845) and rs.syncFrom() (page 991) provide a temporary override of default behavior. If: the mongod (page 1021) instance restarts, the connection to the sync target closes, or Changed in version 2.4: The sync target falls more than 30 seconds behind another member of the replica set; then, the mongod (page 1021) instance will revert to the default sync logic and target.
454
38.1.21 Congure Replica Set Tag Sets

Tag sets let you customize write concern and read preferences for a replica set. MongoDB stores tag sets in the replica set conguration object, which is the document returned by rs.conf() (page 988), in the members[n].tags (page 465) sub-document. This section introduces the conguration of tag sets. For an overview on tag sets and their use, see Replica Set Write Concern (page 399) and Tag Sets (page 404). Differences Between Read Preferences and Write Concerns Custom read preferences and write concerns evaluate tags sets in different ways: Read preferences consider the value of a tag when selecting a member to read from. Write concerns do not use the value of a tag to select a member except to consider whether or not the value is unique. For example, a tag set for a read operation may resemble the following document:
{ "disk": "ssd", "use": "reporting" }
To fulll such a read operation, a member would need to have both of these tags. Any of the following tag sets would satisfy this requirement:
{ { { { "disk": "disk": "disk": "disk": "ssd", "ssd", "ssd", "ssd", "use": "use": "use": "use": "reporting" } "reporting", "rack": 1 } "reporting", "rack": 4 } "reporting", "mem": "64"}
The following tag sets would not be able to fulll this query:
{ { { { { "disk": "ssd" } "use": "reporting" } "disk": "ssd", "use": "production" } "disk": "ssd", "use": "production", "rack": 3 } "disk": "spinning", "use": "reporting", "mem": "32" }
Add Tag Sets to a Replica Set Given the following replica set conguration:
{ "_id" : "rs0", "version" : 1, "members" : [ { "_id" : 0, "host" : "mongodb0.example.net:27017" }, { "_id" : 1, "host" : "mongodb1.example.net:27017" }, { "_id" : 2, "host" : "mongodb2.example.net:27017" }
455
] }
You could add tag sets to the members of this replica set with the following command sequence in the mongo (page 1036) shell:
conf = rs.conf() conf.members[0].tags = { "dc": "east", "use": "production" } conf.members[1].tags = { "dc": "east", "use": "reporting" } conf.members[2].tags = { "use": "production" } rs.reconfig(conf)
After this operation the output of rs.conf() (page 988) would resemble the following:
{ "_id" : "rs0", "version" : 2, "members" : [ { "_id" : 0, "host" : "mongodb0.example.net:27017", "tags" : { "dc": "east", "use": "production" } }, { "_id" : 1, "host" : "mongodb1.example.net:27017", "tags" : { "dc": "east", "use": "reporting" } }, { "_id" : 2, "host" : "mongodb2.example.net:27017", "tags" : { "use": "production" } } ] }
Custom Multi-Datacenter Write Concerns Given a ve member replica set with members in two data centers: 1. a facility VA tagged dc.va 2. a facility GTO tagged dc.gto Create a custom write concern to require conrmation from two data centers using replica set tags, using the following sequence of operations in the mongo (page 1036) shell: 1. Create a replica set conguration JavaScript object conf:
conf = rs.conf()
456
2. Add tags to the replica set members reecting their locations:

conf.members[0].tags conf.members[1].tags conf.members[2].tags conf.members[3].tags conf.members[4].tags rs.reconfig(conf) = = = = = { { { { { "dc.va": "rack1"} "dc.va": "rack2"} "dc.gto": "rack1"} "dc.gto": "rack2"} "dc.va": "rack1"}
3. Create a custom getLastErrorModes (page 466) setting to ensure that the write operation will propagate to at least one member of each facility:
conf.settings = { getLastErrorModes: { MultipleDC : { "dc.va": 1, "dc.gto": 1}}
4. Recongure the replica set using the modied conf conguration object:
rs.reconfig(conf)
To ensure that a write operation propagates to at least one member of the set in both data centers, use the MultipleDC write concern mode as follows:
db.runCommand( { getLastError: 1, w: "MultipleDC" } )
Alternatively, if you want to ensure that each write operation propagates to at least 2 racks in each facility, recongure the replica set as follows in the mongo (page 1036) shell: 1. Create a replica set conguration object conf:
conf = rs.conf()
2. Redene the getLastErrorModes (page 466) value to require two different values of both dc.va and dc.gto:
conf.settings = { getLastErrorModes: { MultipleDC : { "dc.va": 2, "dc.gto": 2}}
rs.reconfig(conf)
Now, the following write concern operation will only return after the write operation propagates to at least two different racks in the each facility:
Congure Tag Sets for Functional Segregation of Read and Write Operations Given a replica set with tag sets that reect: data center facility, physical rack location of instance, and storage system (i.e. disk) type. Where each member of the set has a tag set that resembles one of the following:
{"dc.va": "rack1", disk:"ssd", ssd: "installed" } {"dc.va": "rack2", disk:"raid"} {"dc.gto": "rack1", disk:"ssd", ssd: "installed" }
5
Since read preferences and write concerns use the value of elds in tag sets differently, larger deployments may have some redundancy.
457
{"dc.gto": "rack2", disk:"raid"} {"dc.va": "rack1", disk:"ssd", ssd: "installed" }
To target a read operation to a member of the replica set with a disk type of ssd, you could use the following tag set:
{ disk: "ssd" }
However, to create comparable write concern modes, you would specify a different set of getLastErrorModes (page 466) conguration. Consider the following sequence of operations in the mongo (page 1036) shell: 1. Create a replica set conguration object conf:
conf = rs.conf()
2. Redene the getLastErrorModes (page 466) value to congure two write concern modes:
conf.settings = { "getLastErrorModes" : { "ssd" : { "ssd" : 1 }, "MultipleDC" : { "dc.va" : 1, "dc.gto" : 1 } } }
rs.reconfig(conf)
Now you can specify the MultipleDC write concern mode, as in the following operation, to ensure that a write operation propagates to each data center.
Additionally, you can specify the ssd write concern mode to ensure that a write operation propagates to at least one instance with an SSD.
38.1.22 Recongure a Replica Set with Unavailable Members

To recongure a replica set when a minority of members are unavailable, use the rs.reconfig() (page 989) operation on the current primary, following the example in the Replica Set Reconguration Procedure (page 467). This document provides the following options for re-conguring a replica set when a majority of members are not accessible: Recongure by Forcing the Reconguration (page 458) Recongure by Replacing the Replica Set (page 459) You may need to use one of these procedures, for example, in a geographically distributed replica set, where no local group of members can reach a majority. See Elections (page 389) for more information on this situation. Recongure by Forcing the Reconguration Changed in version 2.0.
458
This procedure lets you recover while a majority of replica set members are down or unreachable. You connect to any surviving member and use the force option to the rs.reconfig() (page 989) method. The force option forces a new conguration onto the. Use this procedure only to recover from catastrophic interruptions. Do not use force every time you recongure. Also, do not use the force option in any automatic scripts and do not use force when there is still a primary. To force reconguration: 1. Back up a surviving member. 2. Connect to a surviving member and save the current conguration. Consider the following example commands for saving the conguration:
cfg = rs.conf() printjson(cfg)
3. On the same member, remove the down and unreachable members of the replica set from the members (page 463) array by setting the array equal to the surviving members alone. Consider the following example, which uses the cfg variable created in the previous step:
cfg.members = [cfg.members[0] , cfg.members[4] , cfg.members[7]]
4. On the same member, recongure the set by using the rs.reconfig() (page 989) command with the force option set to true:
rs.reconfig(cfg, {force : true})
This operation forces the secondary to use the new conguration. The conguration is then propagated to all the surviving members listed in the members array. The replica set then elects a new primary. Note: When you use force : true, the version number in the replica set conguration increases significantly, by tens or hundreds of thousands. This is normal and designed to prevent set version collisions if you accidentally force re-congurations on both sides of a network partition and then the network partitioning ends. 5. If the failure or partition was only temporary, shut down or decommission the removed members as soon as possible. Recongure by Replacing the Replica Set Use the following procedure only for versions of MongoDB prior to version 2.0. If youre running MongoDB 2.0 or later, use the above procedure, Recongure by Forcing the Reconguration (page 458). These procedures are for situations where a majority of the replica set members are down or unreachable. If a majority is running, then skip these procedures and instead use the rs.reconfig() (page 989) command according to the examples in Example Reconguration Operations (page 467). If you run a pre-2.0 version and a majority of your replica set is down, you have the two options described here. Both involve replacing the replica set.
Recongure by Turning Off Replication
This option replaces the replica set with a standalone server. 1. Stop the surviving mongod (page 1021) instances. To ensure a clean shutdown, use an existing control script or an invocation that resembles the following:
459
Set --dbpath (page 1023) to the data directory of your mongod (page 1021) instance. 2. Create a backup of the data directory (i.e. dbpath (page 1081)) of the surviving members of the set. Optional If you have a backup of the database you may instead remove this data. 3. Restart one of the mongod (page 1021) instances without the --replSet (page 1028) parameter. The data is now accessible and provided by a single server that is not a replica set member. Clients can use this server for both reads and writes. When possible, re-deploy a replica set to provide redundancy and to protect your deployment from operational interruption.
Recongure by Breaking the Mirror
This option selects a surviving replica set member to be the new primary and to seed a new replica set. In the following procedure, the new primary is db0.example.net. MongoDB copies the data from db0.example.net to all the other members. 1. Stop the surviving mongod (page 1021) instances. To ensure a clean shutdown, use an existing control script or an invocation that resembles the following:
Set --dbpath (page 1023) to the data directory of your mongod (page 1021) instance. 2. Move the data directories (i.e. dbpath (page 1081)) for all the members except db0.example.net, so that all the members except db0.example.net have empty data directories. For example:
3. Move the data les for local database (i.e. local.*) so that db0.example.net has no local database. For example
mkdir /data/local-old mv /data/db/local* /data/local-old/
4. Start each member of the replica set normally. 5. Connect to db0.example.net in a mongo (page 1036) shell and run rs.initiate() (page 989) to initiate the replica set. 6. Add the other set members using rs.add() (page 987). For example, to add a member running on db1.example.net at port 27017, issue the following command:
rs.add("db1.example.net:27017")
MongoDB performs an initial sync on the added members by copying all data from db0.example.net to the added members. See also: rs.status() (page 990) and db.isMaster() (page 981) Replica Set Reconguration Process (page 467)
460
rs.conf() (page 988) and rs.reconfig() (page 989) Replica Set Conguration (page 463)
461
462
CHAPTER 39
Replica Set Reference Material
Additionally, consider the following reference material for replica sets.
39.1 Replica Set Conguration

39.1.1 Synopsis
This reference provides an overview of replica set conguration options and settings. Use rs.conf() (page 988) in the mongo (page 1036) shell to retrieve this conguration. Note that default values are not explicitly displayed.
39.1.2 Conguration Variables

local.system.replset._id Type: string Value: <setname> An _id eld holding the name of the replica set. This reects the set name congured with replSet (page 1088) or mongod --replSet (page 1028). local.system.replset.members Type: array Contains an array holding an embedded document for each member of the replica set. The members document contains a number of elds that describe the conguration of each member of the replica set. The members (page 463) eld in the replica set conguration document is a zero-indexed array. local.system.replset.members[n]._id Type: ordinal Provides the zero-indexed identier of every member in the replica set. Note: When updating the replica conguration object, address all members of the set using the index value in the array. The array index begins with 0. Do not confuse this index value with the value of the _id (page 463) eld in each document in the members (page 463) array. The _id (page 463) rarely corresponds to the array index.
463
local.system.replset.members[n].host Type: <hostname><:port> Identies the host name of the set member with a hostname and port number. This name must be resolvable for every host in the replica set. Warning: host (page 463) cannot hold a value that resolves to localhost or the local interface unless all members of the set are on hosts that resolve to localhost. local.system.replset.members[n].arbiterOnly Optional. Type: boolean Default: false Identies an arbiter. For arbiters, this value is true, and is automatically congured by rs.addArb() (page 988). local.system.replset.members[n].buildIndexes Optional. Type: boolean Default: true Determines whether the mongod (page 1021) builds indexes on this member. Do not set to false if a replica set can become a master, or if any clients ever issue queries against this instance. Omitting index creation, and thus this setting, may be useful, if: You are only using this instance to perform backups using mongodump (page 1044), this instance will receive no queries, and index creation and maintenance overburdens the host system. If set to false, secondaries congured with this option do build indexes on the _id eld, to facilitate operations required for replication. Warning: You may only set this value when adding a member to a replica set. You may not recongure a replica set to change the value of the buildIndexes (page 464) eld after adding the member to the set. Other secondaries cannot replicate from a members where buildIndexes (page 464) is false. local.system.replset.members[n].hidden Optional. Type: boolean Default: false When this value is true, the replica set hides this instance, and does not include the member in the output of db.isMaster() (page 981) or isMaster (page 838). This prevents read operations (i.e. queries) from ever reaching this host by way of secondary read preference. See also: Hidden Replica Set Members (page 388) local.system.replset.members[n].priority Optional. Type: Number, between 0 and 100.0 including decimals.
464
Chapter 39. Replica Set Reference Material
Default: 1 Specify higher values to make a member more eligible to become primary, and lower values to make the member less eligible to become primary. Priorities are only used in comparison to each other. Members of the set will veto election requests from members when another eligible member has a higher priority value. Changing the balance of priority in a replica set will trigger an election. A priority (page 464) of 0 makes it impossible for a member to become primary. See also: Replica Set Member Priority (page 390) and Replica Set Elections (page 389). local.system.replset.members[n].tags Optional. Type: MongoDB Document Default: none Used to represent arbitrary values for describing or tagging members for the purposes of extending write concern to allow congurable data center awareness. Use in conjunction with getLastErrorModes (page 466) and getLastErrorDefaults (page 466) and db.getLastError() (page 978) (i.e. getLastError (page 831).) For procedures on conguring tag sets, see Congure Replica Set Tag Sets (page 455). local.system.replset.members[n].slaveDelay Optional. Type: Integer. (seconds.) Default: 0 Describes the number of seconds behind the primary that this replica set member should lag. Use this option to create delayed members (page 388), that maintain a copy of the data that reects the state of the data set at some amount of time in the past, specied in seconds. Typically such delayed members help protect against human error, and provide some measure of insurance against the unforeseen consequences of changes and updates. local.system.replset.members[n].votes Optional. Type: Integer Default: 1 Controls the number of votes a server will cast in a replica set election (page 389). The number of votes each member has can be any non-negative integer, but it is highly recommended each member has 1 or 0 votes. If you need more than 7 members in one replica set, use this setting to add additional non-voting members with a votes (page 465) value of 0. For most deployments and most members, use the default value, 1, for votes (page 465). local.system.replset.settings Optional. Type: MongoDB Document The settings document congures options that apply to the whole replica set. local.system.replset.settings.chainingAllowed Optional.
39.1. Replica Set Conguration
465
Type: boolean Default: true New in version 2.2.2. When chainingAllowed (page 465) is true, the replica set allows secondary members to replicate from other secondary members. When chainingAllowed (page 465) is false, secondaries can replicate only from the primary. When you run rs.config() (page 988) to view a replica sets conguration, the chainingAllowed (page 465) eld appears only when set to false. If not set, chainingAllowed (page 465) is true. See also: Manage Chained Replication (page 439) local.system.replset.settings.getLastErrorDefaults Optional. Type: MongoDB Document Specify arguments to the getLastError (page 831) that members of this replica set will use when no arguments to getLastError (page 831) has no arguments. If you specify any arguments, getLastError (page 831) , ignores these defaults. local.system.replset.settings.getLastErrorModes Optional. Type: MongoDB Document Denes the names and combination of members (page 463) for use by the application layer to guarantee write concern to database using the getLastError (page 831) command to provide data-center awareness.
39.1.3 Example Document

The following document provides a representation of a replica set conguration document. Angle brackets (e.g. < and >) enclose all optional elds.
{ _id : <setname>, version: <int>, members: [ { _id : <ordinal>, host : hostname<:port>, <arbiterOnly : <boolean>,> <buildIndexes : <boolean>,> <hidden : <boolean>,> <priority: <priority>,> <tags: { <document> },> <slaveDelay : <number>,> <votes : <number>> } , ... ], <settings: { <getLastErrorDefaults : <lasterrdefaults>,> <chainingAllowed : <boolean>,> <getLastErrorModes : <modes>>
466
}> }
39.1.4 Example Reconguration Operations

Most modications of replica set conguration use the mongo (page 1036) shell. Consider the following reconguration operation: Example Given the following replica set conguration:
{ "_id" : "rs0", "version" : 1, "members" : [ { "_id" : 0, "host" : "mongodb0.example.net:27017" }, { "_id" : 1, "host" : "mongodb1.example.net:27017" }, { "_id" : 2, "host" : "mongodb2.example.net:27017" } ] }
And the following reconguration operation:

cfg = rs.conf() cfg.members[0].priority = 0.5 cfg.members[1].priority = 2 cfg.members[2].priority = 2 rs.reconfig(cfg)
This operation begins by saving the current replica set conguration to the local variable cfg using the rs.conf() (page 988) method. Then it adds priority values to the cfg document where for the rst three sub-documents in the members (page 463) array. Finally, it calls the rs.reconfig() (page 989) method with the argument of cfg to initialize this new conguration. The replica set conguration after this operation will resemble the following:
{ "_id" : "rs0", "version" : 1, "members" : [ { "_id" : 0, "host" : "mongodb0.example.net:27017", "priority" : 0.5 }, { "_id" : 1, "host" : "mongodb1.example.net:27017", "priority" : 2 },
39.1. Replica Set Conguration
467
{ "_id" : 2, "host" : "mongodb2.example.net:27017", "priority" : 1 } ] }
Using the dot notation demonstrated in the above example, you can modify any existing setting or specify any of optional replica set conguration variables (page 463). Until you run rs.reconfig(cfg) at the shell, no changes will take effect. You can issue cfg = rs.conf() at any time before using rs.reconfig() (page 989) to undo your changes and start from the current conguration. If you issue cfg as an operation at any point, the mongo (page 1036) shell at any point will output the complete document with modications for your review. The rs.reconfig() (page 989) operation has a force option, to make it possible to recongure a replica set if a majority of the replica set is not visible, and there is no primary member of the set. use the following form:
rs.reconfig(cfg, { force: true } )
Warning: Forcing a rs.reconfig() (page 989) can lead to rollback situations and other difcult to recover from situations. Exercise caution when using this option.
Note: The rs.reconfig() (page 989) shell method can force the current primary to step down and triggers an election in some situations. When the primary steps down, all clients will disconnect. This is by design. Since this typically takes 10-20 seconds, attempt to make such changes during scheduled maintenance periods.
39.2 Replica Set Commands

This reference collects documentation for all JavaScript methods (page 468) for the mongo (page 1036) shell that support replica set functionality, as well as all database commands (page 472) related to replication function. See Replication (page 385), for a list of all replica set documentation.
39.2.1 JavaScript Methods

The following methods apply to replica sets. For a complete list of all methods, see mongo Shell Methods (page 917). rs.status() Returns A document with status information. This output reects the current status of the replica set, using data derived from the heartbeat packets sent by the other members of the replica set. This method provides a wrapper around the replSetGetStatus (page 840) database command. db.isMaster() Returns A document that describes the role of the mongod (page 1021) instance. If the mongod (page 1021) is a member of a replica set, then the ismaster (page 839) and secondary (page 839) elds report if the instance is the primary or if it is a secondary member of the replica set. See also:
468
isMaster (page 838) for the complete documentation of the output of isMaster() (page 981). rs.initiate(conguration) Parameters conguration Optional. A document that species the conguration of a replica set. If not specied, MongoDB will use a default conguration. Initiates a replica set. Optionally takes a conguration argument in the form of a document that holds the conguration of a replica set. Consider the following model of the most basic conguration for a 3-member replica set:
{ _id : <setname>, members : [ {_id : 0, host : <host0>}, {_id : 1, host : <host1>}, {_id : 2, host : <host2>}, ] }
This function provides a wrapper around the replSetInitiate (page 842) database command. rs.conf() Returns a document that contains the current replica set conguration object. rs.config() rs.config() (page 988) is an alias of rs.conf() (page 988). rs.reconfig(conguration[, force ]) Parameters conguration A document that species the conguration of a replica set. force Optional. Specify { force: true } as the force parameter to force the replica set to accept the new conguration even if a majority of the members are not accessible. Use with caution, as this can lead to rollback situations. Initializes a new replica set conguration. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. rs.reconfig() (page 989) provides a wrapper around the replSetReconfig (page 844) database command. rs.reconfig() (page 989) overwrites the existing replica set conguration. Retrieve the current conguration object with rs.conf() (page 988), modify the conguration as needed and then use rs.reconfig() (page 989) to submit the modied conguration object. To recongure a replica set, use the following sequence of operations:
conf = rs.conf() // modify conf to change configuration rs.reconfig(conf)
If you want to force the reconguration if a majority of the set isnt connected to the current member, or youre issuing the command against a secondary, use the following form:
39.2. Replica Set Commands
469
conf = rs.conf() // modify conf to change configuration rs.reconfig(conf, { force: true } )
Warning: Forcing a rs.reconfig() (page 989) can lead to rollback situations and other difcult to recover from situations. Exercise caution when using this option. See also: Replica Set Conguration (page 463) and Replica Set Administration (page 419).
39.2.2 Denition
rs.add(host, arbiterOnly) Adds a member to a replica set. param string,document host Either a string or a document. If a string, species a host (and optionally port-number) for a new host member for the replica set; MongoDB will add this host with the default conguration. If a document, species any attributes about a member of a replica set. param boolean arbiterOnly If true, this host is an arbiter. If the second argument evaluates to true, as is the case with some documents, then this instance will become an arbiter. You may specify new hosts in one of two ways: 1.as a hostname with an optional port number to use the default conguration as in the Add a Member to an Existing Replica Set (page 425) example. 2.as a conguration document, as in the Add a Member to an Existing Replica Set (Alternate Procedure) (page 426) example. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. rs.add() (page 987) provides a wrapper around some of the functionality of the replSetReconfig (page 844) database command and the corresponding shell helper rs.reconfig() (page 989). See the Replica Set Conguration (page 463) document for full documentation of all replica set conguration options.
39.2.3 Example
To add a mongod (page 1021) accessible on the default port 27017 mongodb3.example.net, use the following rs.add() (page 987) invocation:
rs.add(mongodb3.example.net:27017)
running
on
the
host
If mongodb3.example.net is an arbiter, use the following form:

rs.add(mongodb3.example.net:27017, true)
To add mongodb3.example.net as a secondary-only (page 388) member of set, use the following form of rs.add() (page 987):
rs.add( { "_id": "3", "host": "mongodbd3.example.net:27017", "priority": 0 } )
470
Replace, 3 with the next unused _id value in the replica set. See rs.conf() (page 988) to see the existing _id values in the replica set conguration document. See the Replica Set Conguration (page 463) and Replica Set Administration (page 419) documents for more information. rs.addArb(hostname) Parameters host (string) Species a host (and optionally port-number) for an arbiter member for the replica set. Adds a new arbiter to an existing replica set. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. rs.stepDown(seconds) Parameters seconds (init) Specify the duration of this operation. If not specied the command uses the default value of 60 seconds. Returns disconnects shell. Forces the current replica set member to step down as primary and then attempt to avoid election as primary for the designated number of seconds. Produces an error if the current node is not primary. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. rs.stepDown() (page 991) provides a wrapper around the database command replSetStepDown (page 844). rs.freeze(seconds) Parameters seconds (init) Specify the duration of this operation. Forces the current node to become ineligible to become primary for the period specied. rs.freeze() (page 988) provides a wrapper around the database command replSetFreeze (page 840). rs.remove(hostname) Parameters hostname Specify one of the existing hosts to remove from the current replica set. Removes the node described by the hostname parameter from the current replica set. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. Note: Before running the rs.remove() (page 990) operation, you must shut down the replica set member that youre removing. Changed in version 2.2: This procedure is no longer required when using rs.remove() (page 990), but it remains good practice. rs.slaveOk() Provides a shorthand for the following operation:
471
db.getMongo().setSlaveOk()
This allows the current connection to allow read operations to run on secondary nodes. See the readPref() (page 963) method for more ne-grained control over read preference (page 402) in the mongo (page 1036) shell. db.isMaster() Returns A document that describes the role of the mongod (page 1021) instance. If the mongod (page 1021) is a member of a replica set, then the ismaster (page 839) and secondary (page 839) elds report if the instance is the primary or if it is a secondary member of the replica set. See also: isMaster (page 838) for the complete documentation of the output of isMaster() (page 981). rs.help() Returns a basic help text for all of the replication (page 387) related shell functions. rs.syncFrom() New in version 2.2. Provides a wrapper around the replSetSyncFrom (page 845), which allows administrators to congure the member of a replica set that the current member will pull data from. Specify the name of the member you want to replicate from in the form of [hostname]:[port]. See replSetSyncFrom (page 845) for more details.
39.2.4 Database Commands

The following commands apply to replica sets. For a complete list of all commands, see Database Commands (page 807).
39.2.5 Denition
isMaster isMaster (page 838) returns a document that describes the role of the mongod (page 1021) instance. If the instance is a member of a replica set, then isMaster (page 838) returns a subset of the replica set conguration and status including whether or not the instance is the primary of the replica set. When sent to a mongod (page 1021) instance that is not a member of a replica set, isMaster (page 838) returns a subset of this information. MongoDB drivers and clients use isMaster (page 838) to determine the state of the replica set members and to discover additional members of a replica set. The db.isMaster() (page 981) method in the mongo (page 1036) shell provides a wrapper around isMaster (page 838). The command takes the following form:
{ isMaster: 1 }
See also: db.isMaster() (page 981)
472
39.2.6 Output
All Instances The following isMaster (page 838) elds are common across all roles: isMaster.ismaster A boolean value that reports when this node is writable. If true, then this instance is a primary in a replica set, or a master in a master-slave conguration, or a mongos (page 1032) instance, or a standalone mongod (page 1021). This eld will be false if the instance is a secondary member of a replica set or if the member is an arbiter of a replica set. isMaster.maxBsonObjectSize The maximum permitted size of a BSON object in bytes for this mongod (page 1021) process. If not provided, clients should assume a max size of 4 * 1024 * 1024. isMaster.maxMessageSizeBytes New in version 2.4. The maximum permitted size of a BSON wire protocol message. The default value is 48000000 bytes. isMaster.localTime New in version 2.2. Returns the local server time in UTC. This value is an ISO date. Sharded Instances mongos (page 1032) instances add the following eld to the isMaster (page 838) response document: isMaster.msg Contains the value isdbgrid when isMaster (page 838) returns from a mongos (page 1032) instance. Replica Sets isMaster (page 838) contains these elds when returned by a member of a replica set: isMaster.setName The name of the current :replica set. isMaster.secondary A boolean value that, when true, indicates if the mongod (page 1021) is a secondary member of a replica set. isMaster.hosts An array of strings in the format of "[hostname]:[port]" that lists all members of the replica set that are neither hidden, passive, nor arbiters. Drivers use this array and the isMaster.passives (page 839) to determine which members to read from. isMaster.passives An array of strings in the format of "[hostname]:[port]" listing all members of the replica set which have a priority (page 390) of 0. This eld only appears if there is at least one member with a priority (page 390) of 0. Drivers use this array and the isMaster.hosts (page 839) to determine which members to read from.
473
isMaster.arbiters An array of strings in the format of "[hostname]:[port]" listing all members of the replica set that are arbiters. This eld only appears if there is at least one arbiter in the replica set. isMaster.primary A string in the format of "[hostname]:[port]" listing the current primary member of the replica set. isMaster.arbiterOnly A boolean value that , when true, indicates that the current instance is an arbiter. The arbiterOnly (page 839) eld is only present, if the instance is an arbiter. isMaster.passive A boolean value that, when true, indicates that the current instance is hidden. The passive (page 840) eld is only present for hidden members. isMaster.hidden A boolean value that, when true, indicates that the current instance is hidden. The hidden (page 840) eld is only present for hidden members. isMaster.tags A document that lists any tags assigned to this member. This eld is only present if there are tags assigned to the member. See Congure Replica Set Tag Sets (page 455) for more information. isMaster.me The [hostname]:[port] of the member that returned isMaster (page 838). resync The resync (page 846) command forces an out-of-date slave mongod (page 1021) instance to re-synchronize itself. Note that this command is relevant to master-slave replication only. It does not apply to replica sets. Warning: This command obtains a global write lock and will block other operations until it has completed. replSetFreeze The replSetFreeze (page 840) command prevents a replica set member from seeking election for the specied number of seconds. Use this command in conjunction with the replSetStepDown (page 844) command to make a different node in the replica set a primary. The replSetFreeze (page 840) command uses the following syntax:
{ replSetFreeze: <seconds> }
If you want to unfreeze a replica set member before the specied number of seconds has elapsed, you can issue the command with a seconds value of 0:
{ replSetFreeze: 0 }
Restarting the mongod (page 1021) process also unfreezes a replica set member. replSetFreeze (page 840) is an administrative command, and you must issue it against the admin database.
39.2.7 Denition
replSetGetStatus The replSetGetStatus command returns the status of the replica set from the point of view of the current server. You must run the command against the admin database. The command has the following prototype format:
474
{ replSetGetStatus: 1 }
The value specied does not affect the output of the command. Data provided by this command derives from data included in heartbeats sent to the current instance by other members of the replica set. Because of the frequency of heartbeats, these data can be several seconds out of date. You can also access this functionality through the rs.status() (page 990) helper in the mongo (page 1036) shell. The mongod (page 1021) must have replication enabled and be a member of a replica set for the for replSetGetStatus (page 840) to return successfully.
39.2.8 Output
replSetGetStatus.set The set value is the name of the replica set, congured in the replSet (page 1088) setting. This is the same value as _id (page 463) in rs.conf() (page 988). replSetGetStatus.date The value of the date eld is an ISODate of the current time, according to the current server. Compare this to the value of the lastHeartbeat (page 842) to nd the operational lag between the current host and the other hosts in the set. replSetGetStatus.myState The value of myState (page 841) reects state of the current replica set member. An integer between 0 and 10 represents the state of the member. These integers map to states, as described in the following table: Number 0 1 2 3 4 5 6 7 8 9 10 Name STARTUP PRIMARY SECONDARY RECOVERING FATAL STARTUP2 UNKNOWN ARBITER DOWN ROLLBACK SHUNNED State Start up, phase 1 (parsing conguration.) Primary. Secondary. Member is recovering (initial sync, post-rollback, stale members.) Member has encountered an unrecoverable error. Start up, phase 2 (forking threads.) Unknown (the set has never connected to the member.) Member is an arbiter. Member is not accessible to the set. Member is rolling back data. See rollback. Member has been removed from replica set.
replSetGetStatus.members The members eld holds an array that contains a document for every member in the replica set. replSetGetStatus.members.name The name eld holds the name of the server. replSetGetStatus.members.self The self eld is only included in the document for the current mongod instance in the members array. Its value is true. replSetGetStatus.members.errmsg This eld contains the most recent error or status message received from the member. This eld may be empty (e.g. "") in some cases. replSetGetStatus.members.health The health value is only present for the other members of the replica set (i.e. not the member that returns rs.status (page 990).) This eld conveys if the member is up (i.e. 1) or down (i.e. 0.)
475
replSetGetStatus.members.state The value of the state (page 841) reects state of this replica set member. An integer between 0 and 10 represents the state of the member. These integers map to states, as described in the following table: Number 0 1 2 3 4 5 6 7 8 9 10 Name STARTUP PRIMARY SECONDARY RECOVERING FATAL STARTUP2 UNKNOWN ARBITER DOWN ROLLBACK SHUNNED State Start up, phase 1 (parsing conguration.) Primary. Secondary. Member is recovering (initial sync, post-rollback, stale members.) Member has encountered an unrecoverable error. Start up, phase 2 (forking threads.) Unknown (the set has never connected to the member.) Member is an arbiter. Member is not accessible to the set. Member is rolling back data. See rollback. Member has been removed from replica set.
replSetGetStatus.members.stateStr A string that describes state (page 841). replSetGetStatus.members.uptime The uptime (page 842) eld holds a value that reects the number of seconds that this member has been online. This value does not appear for the member that returns the rs.status() (page 990) data. replSetGetStatus.members.optime A document that contains information regarding the last operation from the operation log that this member has applied. replSetGetStatus.members.optime.t A 32-bit timestamp of the last operation applied to this member of the replica set from the oplog. replSetGetStatus.members.optime.i An incremented eld, which reects the number of operations in since the last time stamp. This value only increases if there is more than one operation per second. replSetGetStatus.members.optimeDate An ISODate formatted date string that reects the last entry from the oplog that this member applied. If this differs signicantly from lastHeartbeat (page 842) this member is either experiencing replication lag or there have not been any new operations since the last update. Compare members.optimeDate between all of the members of the set. replSetGetStatus.members.lastHeartbeat The lastHeartbeat value provides an ISODate formatted date of the last heartbeat received from this member. Compare this value to the value of the date (page 841) eld to track latency between these members. This value does not appear for the member that returns the rs.status() (page 990) data. replSetGetStatus.members.pingMS The pingMS represents the number of milliseconds (ms) that a round-trip packet takes to travel between the remote member and the local instance. This value does not appear for the member that returns the rs.status() (page 990) data. replSetGetStatus.syncingTo The syncingTo eld is only present on the output of rs.status() (page 990) on secondary and recovering members, and holds the hostname of the member from which this instance is syncing.
476
replSetInitiate The replSetInitiate (page 842) command initializes a new replica set. Use the following syntax:
{ replSetInitiate : <config_document> }
The <config_document> is a document that species the replica sets conguration. For instance, heres a cong document for creating a simple 3-member replica set:
A typical way of running this command is to assign the cong document to a variable and then to pass the document to the rs.initiate() (page 989) helper:
config = { _id : "my_replica_set", members : [ {_id : 0, host : "rs1.example.net:27017"}, {_id : 1, host : "rs2.example.net:27017"}, {_id : 2, host : "rs3.example.net", arbiterOnly: true}, ] } rs.initiate(config)
Notice that omitting the port cause the host to use the default port of 27017. Notice also that you can specify other options in the cong documents such as the arbiterOnly setting in this example. See also: Replica Set Conguration (page 463), Replica Set Administration (page 419), and Replica Set Reconguration (page 467). replSetMaintenance The replSetMaintenance (page 843) admin command enables or disables the maintenance mode for a secondary member of a replica set. The command has the following prototype form:
{ replSetMaintenance: <boolean> }
Consider the following behavior when running the replSetMaintenance (page 843) command: You cannot run the command on the Primary. You must run the command against the admin database. When enabled replSetMaintenance: secondary is RECOVERING: 1, the member enters the RECOVERING state. While the
The member is not accessible for read operations. The member continues to sync its oplog from the Primary. replSetReconfig The replSetReconfig (page 844) command modies the conguration of an existing replica set. You can
477
use this command to add and remove members, and to alter the options set on existing members. Use the following syntax:
{ replSetReconfig: <new_config_document>, force: false }
You may also run the command using the shells rs.reconfig() (page 989) method. Be aware of the following replSetReconfig (page 844) behaviors: You must issue this command against the admin database of the current primary member of the replica set. You can optionally force the replica set to accept the new conguration by specifying force: true. Use this option if the current member is not primary or if a majority of the members of the set are not accessible. Warning: Forcing the replSetReconfig (page 844) command can lead to a rollback situation. Use with caution. Use the force option to restore a replica set to new servers with different hostnames. This works even if the set members already have a copy of the data. A majority of the sets members must be operational for the changes to propagate properly. This command can cause downtime as the set renegotiates primary-status. Typically this is 10-20 seconds, but could be as long as a minute or more. Therefore, you should attempt to recongure only during scheduled maintenance periods. In some cases, replSetReconfig (page 844) forces the current primary to step down, initiating an election for primary among the members of the replica set. When this happens, the set will drop all current connections. Note: replSetReconfig (page 844) obtains a special mutually exclusive lock to prevent more than one replSetReconfig (page 844) operation from occurring at the same time. replSetSyncFrom New in version 2.2. Options host Species the name and port number of the replica set member that this member replicates from. Use the [hostname]:[port] form. replSetSyncFrom (page 845) allows you to explicitly congure which host the current mongod (page 1021) will poll oplog entries from. This operation may be useful for testing different patterns and in situations where a set member is not replicating from the host you want. The member to replicate from must be a valid source for data in the set. A member cannot replicate from: itself. an arbiter, because arbiters do not hold data. a member that does not build indexes. an unreachable member. a mongod (page 1021) instance that is not a member of the same replica set. If you attempt to replicate from a member that is more than 10 seconds behind the current member, mongod (page 1021) will return and log a warning, but it still will replicate from the member that is behind.
478
If you run rs.syncFrom() (page 991) during initial sync, MongoDB produces no error messages, but the sync target will not change until after the initial sync operation. The command has the following prototype form:
{ replSetSyncFrom: "[hostname]:[port]" }
To run the command in the mongo (page 1036) shell, use the following invocation:
db.adminCommand( { replSetSyncFrom: "[hostname]:[port]" } )
You may also use the rs.syncFrom() (page 991) helper in the mongo (page 1036) shell, in an operation with the following form:
rs.syncFrom("[hostname]:[port]")
Note: replSetSyncFrom (page 845) and rs.syncFrom() (page 991) provide a temporary override of default behavior. If: the mongod (page 1021) instance restarts, the connection to the sync target closes, or Changed in version 2.4: The sync target falls more than 30 seconds behind another member of the replica set; then, the mongod (page 1021) instance will revert to the default sync logic and target.
39.3 Replica Set Features and Version Compatibility

Note: This table is for archival purposes and does not list all features of replica sets. Always use the latest stable release of MongoDB in production deployments. Features Slave Delay Hidden replSetFreeze (page 840) and replSetStepDown (page 844) Replicated ops in mongostat (page 1063) Syncing from Secondaries Authentication Replication from Nearest Server (by ping Time) replSetSyncFrom (page 845) support for replicating from specic members. Additionally: 1.8-series secondaries can replicate from 1.6-series primaries. 1.6-series secondaries cannot replicate from 1.8-series primaries. See also: http://docs.mongodb.org/manual/reference/command/replSetGetStatus db.getReplicationInfo() (page 979). and Version 1.6.3 1.7 1.7.3 1.7.3 1.8.0 1.8.0 2.0 2.2
39.3. Replica Set Features and Version Compatibility
479
480
Part IX
Sharding
481
Sharding distributes a single logical database system across a cluster of machines.
483
484
CHAPTER 40
Sharding Concepts
40.1 Sharded Cluster Overview

Sharding is MongoDBs approach to scaling out. Sharding partitions a collection and stores the different portions on different machines. When a databases collections become too large for existing storage, you need only add a new machine. Sharding automatically distributes collection data to the new server. Sharding automatically balances data and load across machines. Sharding provides additional write capacity by distributing the write load over a number of mongod (page 1021) instances. Sharding allows users to increase the potential amount of data in the working set.
40.1.1 Sharding
Sharding occurs within a sharded cluster. A sharded cluster consists of the following components: Shards. A shard is a container that holds a subset of a collections data. Each shard is either a single mongod (page 1021) instance or a replica set. In production, all shards should be replica sets. Applications do not access shards directly but instead access mongos instances (page 490). Cong servers. Each cong server (page 500) is a mongod (page 1021) instance that holds metadata about the cluster. The metadata maps chunks to shards. mongos instances. The mongos instances (page 490) route the reads and writes from applications to the shards. Applications do not access the shards directly. Within a sharded cluster, you enable sharding on a per-database basis. When you enable sharding on a database, MongoDB distributes the collections across the shards. MongoDB does not yet distribute data within a collection. After enabling sharding for a database, you choose which collections to shard. For each sharded collection, you specify a shard key. To set up a sharded cluster, see Deploy a Sharded Cluster (page 503).
40.1.2 Shard Keys

The shard key determines the distribution of the collections documents among the clusters shards. The shard key is a eld that exists in every document in the collection. MongoDB distributes documents according to ranges of values in the shard key. A given shard holds documents for which the shard key falls within a specic range of values. Shard keys, like indexes, can be either a single eld or multiple elds.
485
Within a shard, MongoDB further partitions documents into chunks. Each chunk represents a smaller range of values within the shards range. When a chunk grows beyond the chunk size (page 498), MongoDB splits the chunk into smaller chunks, always based on ranges in the shard key. To select a shard key, see Select a Shard Key (page 507).
40.1.3 Hashed Sharding

New in version 2.4. Hashed shard keys (page 496) use a hashed index (page 350) of a single eld as the shard key to partition data across your sharded cluster. The eld you choose as your hashed shard key should have a good cardinality, or large number of different values. Hashed keys work well with elds that increase monotonically like ObjectId values or timestamps. If you shard an empty collection using a hashed shard key, MongoDB will automatically create and migrate chunks so that each shard has two chunks. You can control how many chunks MongoDB will create with the numInitialChunks parameter to shardCollection (page 851) or by manually creating chunks on the empty collection using the split (page 853) command. To shard a collection using a hashed shard key, see Hashed Sharding (page 507).
40.1.4 Shard Balancing

Balancing is the process MongoDB uses to redistribute data within a sharded cluster. When a shard has too many chunks when compared to other shards, MongoDB automatically balances the shards. MongoDB balances the shards without intervention from the application layer. The balancing process attempts to minimize the impact that balancing can have on the cluster, by: Moving only one chunk at a time. Initiating a balancing round only when the difference in the number of chunks between the shard with the greatest number and the shard with the lowest exceeds the migration threshold (page 497). You may disable the balancer on a temporary basis for maintenance and limit the window during which it runs to prevent the balancing process from impacting production trafc. To disable the balancer, see Disable the Balancer (page 529). See also: Manage Sharded Cluster Balancer (page 527). Note: The balancing procedure for sharded clusters is entirely transparent to the user and application layer. This documentation is only included for your edication and possible troubleshooting purposes.
40.1.5 When to Use Sharding

While sharding is a powerful and compelling feature, it comes with signicant infrastructure requirements and some limited complexity costs. As a result, use sharding only as necessary and when indicated by actual operational requirements. You should consider deploying a sharded cluster if: your data set approaches or exceeds the storage capacity of a single node in your system.
486
Chapter 40. Sharding Concepts
the size of your systems active working set will soon exceed the capacity of the maximum amount of RAM for your system. your system has a large amount of write activity, a single MongoDB instance cannot write data fast enough to meet demand, and all other approaches have not reduced contention. If these attributes are not present in your system, sharding will only add additional complexity to your system without providing much benet. When designing your data model, if you will eventually need a sharded cluster, consider which collections you will want to shard and the corresponding shard keys. Warning: It takes time and resources to deploy sharding, and if your system has already reached or exceeded its capacity, you will have a difcult time deploying sharding without impacting your application. As a result, if you think you will need to partition your database in the future, do not wait until your system is overcapacity to enable sharding. For information on requirements, see the following: Infrastructure Requirements for Sharded Clusters (page 487) Data Quantity Requirements for Sharded Clusters (page 488)
40.2 Sharded Cluster Architectures

This document describes requirements, organization and design of sharded cluster deployments.
40.2.1 Infrastructure Requirements for Sharded Clusters

A sharded cluster has the following components: Three cong servers. These special mongod (page 1021) instances store the metadata for the cluster. The mongos (page 1032) instances cache this data and use it to determine which shard is responsible for which chunk. For development and testing purposes you may deploy a cluster with a single conguration server process, but always use exactly three cong servers for redundancy and safety in production. Two or more shards. Each shard consists of one or more mongod (page 1021) instances that store the data for the shard. These normal mongod (page 1021) instances hold all of the actual data for the cluster. Typically each shard is a replica sets. Each replica set consists of multiple mongod (page 1021) instances. The members of the replica set provide redundancy and high availability for the data in each shard. Warning: MongoDB enables data partitioning, or sharding, on a per collection basis. You must access all data in a sharded cluster via the mongos (page 1032) instances as below. If you connect directly to a mongod (page 1021) in a sharded cluster you will see its fraction of the clusters data. The data on any given shard may be somewhat random: MongoDB provides no guarantee that any two contiguous chunks will reside on a single shard. One or more mongos (page 1032) instances. These instance direct queries from the application layer to the shards that hold the data. The mongos (page 1032) instances have no persistent state or data les and only cache metadata in RAM from the cong servers.
40.2. Sharded Cluster Architectures
487
Note: In most situations mongos (page 1032) instances use minimal resources, and you can run them on your application servers without impacting application performance. However, if you use the aggregation framework some processing may occur on the mongos (page 1032) instances, causing that mongos (page 1032) to require more system resources.
40.2.2 Data Quantity Requirements for Sharded Clusters

Your cluster must manage a signicant quantity of data for sharding to have an effect on your collection. The default chunk size is 64 megabytes, and the balancer (page 486) will not begin moving data until the imbalance of chunks in the cluster exceeds the migration threshold (page 497). Practically, this means that unless your cluster has many hundreds of megabytes of data, chunks will remain on a single shard. While there are some exceptional situations where you may need to shard a small collection of data, most of the time the additional complexity added by sharding the small collection is not worth the additional complexity and overhead unless you need additional concurrency or capacity for some reason. If you have a small data set, usually a properly congured single MongoDB instance or replica set will be more than sufcient for your persistence layer needs. Chunk size is user configurable (page 1034). However, the default value is of 64 megabytes is ideal for most deployments. See the Chunk Size (page 498) section in the Sharded Cluster Internals (page 493) document for more information.
40.2.3 Restriction on the Use of the localhost Interface

Because all components of a sharded cluster must communicate with each other over the network, there are special restrictions regarding the use of localhost addresses: If you use either localhost or 127.0.0.1 as the host identier, then you must use localhost or 127.0.0.1 for all host settings for any MongoDB instances in the cluster. This applies to both the host argument to addShard (page 846) and the value to the mongos --configdb (page 1034) run time option. If you mix localhost addresses with remote host address, MongoDB will produce errors.
40.2.4 Test Cluster Architecture

You can deploy a very minimal cluster for testing and development. These non-production clusters have the following components: One cong server (page 500). At least one mongod (page 1021) instance (either replica sets or as a standalone node.) One mongos (page 1032) instance. Warning: Use the test cluster architecture for testing and development only.
40.2.5 Production Cluster Architecture

In a production cluster, you must ensure that data is redundant and that your systems are highly available. To that end, a production-level cluster must have the following components:
488
Three cong servers (page 500), each residing on a discrete system. A single sharded cluster must have exclusive use of its cong servers (page 500). If you have multiple sharded clusters, you will need to have a group of cong servers for each cluster. Two or more replica sets to serve as shards. For information on replica sets, see Replication (page 385). Two or more mongos (page 1032) instances. Typically, you deploy a single mongos (page 1032) instance on each application server. Alternatively, you may deploy several mongos (page 1032) nodes and let your application connect to these via a load balancer.
40.2.6 Sharded and Non-Sharded Data

Sharding operates on the collection level. You can shard multiple collections within a database, or have multiple databases with sharding enabled. 1 However, in production deployments some databases and collections will use sharding, while other databases and collections will only reside on a single database instance or replica set (i.e. a shard.) Regardless of the data architecture of your sharded cluster, ensure that all queries and operations use the mongos router to access the data cluster. Use the mongos (page 1032) even for operations that do not impact the sharded data. Every database has a primary 2 shard that holds all un-sharded collections in that database. All collections that are not sharded reside on the primary for their database. Use the movePrimary (page 850) command to change the primary shard for a database. Use the db.printShardingStatus() (page 983) command or the sh.status() (page 1002) to see an overview of the cluster, which contains information about the chunk and database distribution within the cluster. Warning: The movePrimary (page 850) command can be expensive because it copies all non-sharded data to the new shard, during which that data will be unavailable for other operations. When you deploy a new sharded cluster, the rst shard becomes the primary for all databases before enabling sharding. Databases created subsequently, may reside on any shard in the cluster.
40.2.7 High Availability and MongoDB

A production (page 488) cluster has no single point of failure. This section introduces the availability concerns for MongoDB deployments and highlights potential failure scenarios and available resolutions: Application servers or mongos (page 1032) instances become unavailable. If each application server has its own mongos (page 1032) instance, other application servers can continue access the database. Furthermore, mongos (page 1032) instances do not maintain persistent state, and they can restart and become unavailable without loosing any state or data. When a mongos (page 1032) instance starts, it retrieves a copy of the cong database and can begin routing queries. A single mongod (page 1021) becomes unavailable in a shard. Replica sets (page 385) provide high availability for shards. If the unavailable mongod (page 1021) is a primary, then the replica set will elect (page 389) a new primary. If the unavailable mongod (page 1021) is a secondary, and it disconnects the primary and secondary will continue to hold all data. In a three member replica set, even if a single member of the set experiences catastrophic failure, two other members have full copies of the data. 3
1 As you congure sharding, you will use the enableSharding (page 848) command to enable sharding for a database. This simply makes it possible to use the shardCollection (page 851) command on a collection within that database. 2 The term primary in the context of databases and sharding, has nothing to do with the term primary in the context of replica sets. 3 If an unavailable secondary becomes available while it still has current oplog entries, it can catch up to the latest state of the set using the normal replication process, otherwise it must perform an initial sync.
40.2. Sharded Cluster Architectures
489
Always investigate availability interruptions and failures. If a system is unrecoverable, replace it and create a new member of the replica set as soon as possible to replace the lost redundancy. All members of a replica set become unavailable. If all members of a replica set within a shard are unavailable, all data held in that shard is unavailable. However, the data on all other shards will remain available, and its possible to read and write data to the other shards. However, your application must be able to deal with partial results, and you should investigate the cause of the interruption and attempt to recover the shard as soon as possible. One or two cong database become unavailable. Three distinct mongod (page 1021) instances provide the cong database using a special two-phase commits to maintain consistent state between these mongod (page 1021) instances. Cluster operation will continue as normal but chunk migration (page 486) and the cluster can create no new chunk splits (page 521). Replace the cong server as soon as possible. If all multiple cong databases become unavailable, the cluster can become inoperable. Note: All cong servers must be running and available when you rst initiate a sharded cluster.
40.3 Query Routing in Sharded Clusters

MongoDB provides the mongos (page 1032) program to handle query routing in a sharded cluster.
40.3.1 mongos Operational Overview

The mongos (page 1032) program provides a single unied interface to a sharded cluster. An application does not access shards (i.e. the mongod (page 1021) instances) directly but instead accesses the mongos (page 1032). The mongos (page 1032) routes queries and returns results. The mongos (page 1032) keeps track of what data is on which shard by caching that information as metadata on the cong servers (page 500). The mongos (page 1032) uses the metadata to route operations from applications and clients to the mongod (page 1021) instances. A mongos (page 1032) has no persistent state and consume minimal system resources. The most common practice is to run mongos (page 1032) instances on the same systems as your application servers, but you can maintain mongos (page 1032) instances on the shards or on other dedicated resources. Note: Changed in version 2.1. Some aggregation operations using the aggregate (page 808) command (i.e. db.collection.aggregate() (page 918),) will cause mongos (page 1032) instances to require more CPU resources than in previous versions. This modied performance prole may dictate alternate architecture decisions if you use the aggregation framework extensively in a sharded environment.
40.3.2 Broadcast Operations vs Targeted Operations

In general, operations in a sharded environment are either: Broadcast to all shards in the cluster that hold documents in a collection Targeted at a single shard or a limited group of shards, based on the shard key
490
For best performance, use targeted operations whenever possible. While some operations must broadcast to all shards, you can ensure MongoDB uses targeted operations whenever possible by always including the shard key. Broadcast Operations A query operation is broadcast to all shards stores this data.
4
unless the mongos (page 1032) can determine which shard or shard
Multi-update operations are always broadcast operations. The remove() (page 944) operation is always a broadcast operation, unless the operation species the shard key in full. Targeted Operations All insert() (page 936) operations target to one shard. All single update() (page 948) operations target to one shard. This includes upsert operations. For queries that include the shard key, mongos (page 1032) can target the query at a specic shard or set of shards. This is the case only if the portion of the shard key included in the query is a prex of the shard key. For example, if the shard key is:
{ a: 1, b: 1, c: 1 }
The mongos (page 1032) program can route queries that include the full shard key or either of the following shard key prexes at a specic shard or set of shards:
{ a: 1 } { a: 1, b: 1 }
Depending on the distribution of data in the cluster and the selectivity of the query, mongos (page 1032) may still have to contact multiple shards 5 to fulll these queries.
40.3.3 mongos Query Routing

A mongos (page 1032) instance uses the following process to route a query to a cluster. The mongos (page 1032): 1. Determines the list of shards that must receive the query. 2. Establishes a cursor on all targeted shards. Continue reading for additional information: mongos Determines which Shards Receive a Query In some cases, when the shard key or a prex of the shard key is a part of the query, the mongos (page 1032) can route the query to a subset of the shards. Otherwise, the mongos (page 1032) must direct the query to all shards that hold documents for that collection. Example Given the following shard key:
4 5
If a shard does not store chunks from a given collection, queries for documents in that collection are not broadcast to that shard. mongos (page 1032) will route some queries, even some that include the shard key, to all shards, if needed.
40.3. Query Routing in Sharded Clusters
491
{ zipcode: 1, u_id: 1, c_date: 1 }
Depending on the distribution of chunks in the cluster, the mongos (page 1032) may be able to target the query at a subset of shards, if the query contains the following elds:
{ zipcode: 1 } { zipcode: 1, u_id: 1 } { zipcode: 1, u_id: 1, c_date: 1 }
mongos Establishes a Cursor on Targeted Shards When the rst batch of results returns from the cursors: For query with sorted results (i.e. using cursor.sort() (page 965)) the mongos (page 1032) instance performs a merge sort of all queries. For a query with unsorted results, the mongos (page 1032) instance returns a result cursor that round robins results from all cursors on the shards. Changed in version 2.0.5: Before 2.0.5, the mongos (page 1032) exhausted each cursor, one by one.
40.4 Security Practices for Sharded Clusters

MongoDB controls access to sharded clusters with key les that store authentication credentials. The components of sharded clusters use the secret stored in the key les when authenticating to each other. Create key les and then point your mongos (page 1032) and mongod (page 1021) instances to the les, as described later in this section. Beyond the auth (page 1081) mechanisms described in this section, always run your sharded clusters in trusted networking environments that limit access to the cluster with network rules. Your networking environments should enforce restrictions that ensure only known trafc reaches your mongos (page 1032) and mongod (page 1021) instances. This section describes authentication specic to sharded clusters. For information on authentication across MongoDB, see Access Control (page 135).
40.4.1 Access Control Privileges in Sharded Clusters

In sharded clusters, MongoDB provides separate administrative privileges for the sharded cluster and for each shard. Beyond these administration privileges, privileges for sharded cluster deployments are functionally the same as any other MongoDB deployment. See, Access Control (page 135) for more information. For sharded clusters, MongoDB provides these separate administrative privileges: Administrative privileges for the sharded cluster. These privileges provide read-and-write access to the cong servers admin. These users can run all administrative commands. Administrative privileges also give the user read-and-write access to all the clusters databases. The credentials for administrative privileges on the cluster reside on the cong servers. To receive admin access to the cluster, you must authenticate a session while connected to a mongos (page 1032) instance using the admin database. Administrative privileges for the mongod (page 1021) instance, or replica set, that provides each individual shard. Each shard has its own admin database that stores administrative credentials and access for that shard only. These credentials are completely distinct from the cluster-wide administrative credentials.
492
For more information on privileges, see Access Control (page 135).
40.4.2 Access a Sharded Cluster with Authentication

To access a sharded cluster as an authenticated user, use the appropriate authentication options in mongo (page 1036). To access a sharded cluster as an authenticated, non-admin user, see either of the following: authenticate db.auth() (page 968) To terminate an authenticated session, see the logout command.
40.5 Sharded Cluster Internals

This document introduces lower level sharding concepts for users who are familiar with sharding generally and want to learn more about the internals. This document provides a more detailed understanding of your clusters behavior. For higher level sharding concepts, see Sharded Cluster Overview (page 485). For complete documentation of sharded clusters see the Sharding (page 483) section of this manual.
40.5.1 Shard Keys

Shard keys are the eld in a collection that MongoDB uses to distribute documents within a sharded cluster. See the overview of shard keys (page 485) for a higher-level introduction. Cardinality Cardinality in the context of MongoDB, refers to the ability of the system to partition data into chunks. For example, consider a collection of data such as an address book that stores address records: Consider the use of a state eld as a shard key: The state keys value holds the US state for a given address document. This eld has a low cardinality as all documents that have the same value in the state eld must reside on the same shard, even if a particular states chunk exceeds the maximum chunk size. Since there are a limited number of possible values for the state eld, MongoDB may distribute data unevenly among a small number of xed chunks. This may have a number of effects: If MongoDB cannot split a chunk because all of its documents have the same shard key, migrations involving these un-splitable chunks will take longer than other migrations, and it will be more difcult for your data to stay balanced. If you have a xed maximum number of chunks, you will never be able to use more than that number of shards for this collection. Consider the use of a zipcode eld as a shard key: While this eld has a large number of possible values, and thus has potentially higher cardinality, its possible that a large number of users could have the same value for the shard key, which would make this chunk of users un-splitable. In these cases, cardinality depends on the data. If your address book stores records for a geographically distributed contact list (e.g. Dry cleaning businesses in America,) then a value like zipcode would be sufcient.
40.5. Sharded Cluster Internals
493
However, if your address book is more geographically concentrated (e.g ice cream stores in Boston Massachusetts,) then you may have a much lower cardinality. Consider the use of a phone-number eld as a shard key: Phone number has a high cardinality, because users will generally have a unique value for this eld, MongoDB will be able to split as many chunks as needed. While high cardinality, is necessary for ensuring an even distribution of data, having a high cardinality does not guarantee sufcient query isolation (page 494) or appropriate write scaling (page 494). Please continue reading for additional information. Write Scaling Some possible shard keys will allow your application to take advantage of the increased write capacity that the cluster can provide, while others do not. Consider the following example where you shard by the values of the default _id eld, which is ObjectID. ObjectID is computed upon document creation, that is a unique identier for the object. However, the most significant bits of data in this value represent a time stamp, which means that they increment in a regular and predictable pattern. Even though this value has high cardinality (page 493), when using this, any date, or other monotonically increasing number as the shard key, all insert operations will be storing data into a single chunk, and therefore, a single shard. As a result, the write capacity of this shard will dene the effective write capacity of the cluster. A shard key that increases monotonically will not hinder performance if you have a very low insert rate, or if most of your write operations are update() (page 948) operations distributed through your entire data set. Generally, choose shard keys that have both high cardinality and will distribute write operations across the entire cluster. Typically, a computed shard key that has some amount of randomness, such as ones that include a cryptographic hash (i.e. MD5 or SHA1) of other content in the document, will allow the cluster to scale write operations. However, random shard keys do not typically provide query isolation (page 494), which is another important characteristic of shard keys. Querying The mongos (page 1032) provides an interface for applications to interact with sharded clusters that hides the complexity of data partitioning. A mongos (page 1032) receives queries from applications, and uses metadata from the cong server (page 500), to route queries to the mongod (page 1021) instances with the appropriate data. While the mongos (page 1032) succeeds in making all querying operational in sharded environments, the shard key you select can have a profound affect on query performance. See also: The mongos and Sharding (page 490) and cong server (page 500) sections for a more general overview of querying in sharded environments.
Query Isolation
The fastest queries in a sharded environment are those that mongos (page 1032) will route to a single shard, using the shard key and the cluster meta data from the cong server (page 500). For queries that dont include the shard key, mongos (page 1032) must query all shards, wait for their response and then return the result to the application. These scatter/gather queries can be long running operations.
494
If your query includes the rst component of a compound shard key 6 , the mongos (page 1032) can route the query directly to a single shard, or a small number of shards, which provides better performance. Even if you query values of the shard key reside in different chunks, the mongos (page 1032) will route queries directly to specic shards. To select a shard key for a collection: determine the most commonly included elds in queries for a given application nd which of these operations are most performance dependent. If this eld has low cardinality (i.e not sufciently selective) you should add a second eld to the shard key making a compound shard key. The data may become more splitable with a compound shard key. See also: mongos Operational Overview (page 490) for more information on query operations in the context of sharded clusters. Specically the mongos Operational Overview (page 490) sub-section outlines the procedure that mongos (page 1032) uses to route read operations to the shards.
Sorting
In sharded systems, the mongos (page 1032) performs a merge-sort of all sorted query results from the shards. See the sharded query routing (page 490) and Use Indexes to Sort Query Results (page 343) sections for more information. Operations and Reliability The most important consideration when choosing a shard key are: to ensure that MongoDB will be able to distribute data evenly among shards, and to scale writes across the cluster, and to ensure that mongos (page 1032) can isolate most queries to a specic mongod (page 1021). Furthermore: Each shard should be a replica set, if a specic mongod (page 1021) instance fails, the replica set members will elect another to be primary and continue operation. However, if an entire shard is unreachable or fails for some reason, that data will be unavailable. If the shard key allows the mongos (page 1032) to isolate most operations to a single shard, then the failure of a single shard will only render some data unavailable. If your shard key distributes data required for every operation throughout the cluster, then the failure of the entire shard will render the entire cluster unavailable. In essence, this concern for reliability simply underscores the importance of choosing a shard key that isolates query operations to a single shard. Choosing a Shard Key For many data sets, there may be no single, naturally occurring key in your collection that possesses all of the qualities of a good shard key. For these cases, you may select one of the following strategies: 1. Compute a more ideal shard key in your application layer, and store this in all of your documents, potentially in the _id eld.
In many ways, you can think of the shard key a cluster-wide unique index. However, be aware that sharded systems cannot enforce cluster-wide unique indexes unless the unique eld is in the shard key. Consider the Indexing Overview (page 329) page for more information on indexes and compound indexes.
6
495
2. Use a compound shard key that uses two or three values from all documents that provide the right mix of cardinality with scalable write operations and query isolation. 3. Determine that the impact of using a less than ideal shard key, is insignicant in your use case given: limited write volume, expected data size, or query patterns and demands. 4. New in version 2.4: Use a hashed shard key. With a hashed shard key, you can choose a eld that has high cardinality and create a hashed indexes (page 350) index on that eld. MongoDB then uses the values of this hashed index as the shard key values, thus ensuring an even distribution across the shards. From a decision making stand point, begin by nding the eld that will provide the required query isolation (page 494), ensure that writes will scale across the cluster (page 494), and then add an additional eld to provide additional cardinality (page 493) if your primary key does not have sufcient split-ability. Shard Key Indexes All sharded collections must have an index that starts with the shard key. If you shard a collection that does not yet contain documents and without such an index, the shardCollection (page 851) command will create an index on the shard key. If the collection already contains documents, you must create an appropriate index before using shardCollection (page 851). Changed in version 2.2: The index on the shard key no longer needs to be identical to the shard key. This index can be an index of the shard key itself as before, or a compound index where the shard key is the prex of the index. This index cannot be a multikey index. If you have a collection named people, sharded using the eld { zipcode: with an index on the eld { zipcode: 1, username: 1 }, then: 1. Create an index on { zipcode: 1, username: 1 }: 1 }, and you want to replace this
db.people.ensureIndex( { zipcode: 1, username: 1 } );
2. When MongoDB nishes building the index, you can safely drop existing index on { zipcode:
db.people.dropIndex( { zipcode: 1 } );
1 }:
Warning: The index on the shard key cannot be a multikey index. As above, an index on { zipcode: 1, username: 1 } can only replace an index on zipcode if there are no array values for the username eld. If you drop the last appropriate index for the shard key, recover by recreating an index on just the shard key.
Hashed Shard Keys New in version 2.4. Hashed shard keys use a special hashed index type (page 335) to store hashes of the shard key eld to partition data in a cluster. Use hashed shard keys when you want to shard using a eld that increases monotonically, like an ObjectId, or has high cardinality but uneven distribution. Example
496
A hashed index on an ObjectId will lead to an even distribution of documents across all shards since the hash of two sequential documents will have different hashes. Note: Hash-based sharding does not support tag-aware sharding. Warning: hashed indexes truncate oating point numbers to 64-bit integers before hashing. For example, a hashed index would store the same value for a eld that held a value of 2.3, 2.2 and 2.9. To prevent collisions, do not use a hashed index for oating point numbers that cannot be consistently converted to 64-bit integers (and then back to oating point.) hashed indexes do not support oating point values larger than 253 .
40.5.2 Cluster Balancer

The balancer (page 486) sub-process is responsible for redistributing chunks evenly among the shards and ensuring that each member of the cluster is responsible for the same volume of data. This section contains complete documentation of the balancer process and operations. For a higher level introduction see the Shard Balancing (page 486) section. Balancing Internals A balancing round originates from an arbitrary mongos (page 1032) instance from one of the clusters mongos (page 1032) instances. When a balancer process is active, the responsible mongos (page 1032) acquires a lock by modifying a document in the lock collection in the Cong Database (page 545). By default, the balancer process is always running. When the number of chunks in a collection is unevenly distributed among the shards, the balancer begins migrating chunks from shards with more chunks to shards with a fewer number of chunks. The balancer will continue migrating chunks, one at a time, until the data is evenly distributed among the shards. While these automatic chunk migrations are crucial for distributing data, they carry some overhead in terms of bandwidth and workload, both of which can impact database performance. As a result, MongoDB attempts to minimize the effect of balancing by only migrating chunks when the distribution of chunks passes the migration thresholds (page 497). The migration process ensures consistency and maximizes availability of chunks during balancing: when MongoDB begins migrating a chunk, the database begins copying the data to the new server and tracks incoming write operations. After migrating chunks, the from mongod (page 1021) sends all new writes to the receiving server. Finally, mongos (page 1032) updates the chunk record in the cong database to reect the new location of the chunk. Note: Changed in version 2.0: Before MongoDB version 2.0, large differences in timekeeping (i.e. clock skew) between mongos (page 1032) instances could lead to failed distributed locks, which carries the possibility of data loss, particularly with skews larger than 5 minutes. Always use the network time protocol (NTP) by running ntpd on your servers to minimize clock skew.
Migration Thresholds Changed in version 2.2: The following thresholds appear rst in 2.2; prior to this release, balancing would only commence if the shard with the most chunks had 8 more chunks than the shard with the least number of chunks. In order to minimize the impact of balancing on the cluster, the balancer will not begin balancing until the distribution of chunks has reached certain thresholds. These thresholds apply to the difference in number of chunks between
497
the shard with the greatest number of chunks and the shard with the least number of chunks. The balancer has the following thresholds: Number of Chunks Less than 20 21-80 Greater than 80 Migration Threshold 2 4 8
Once a balancing round starts, the balancer will not stop until the difference between the number of chunks on any two shards is less than two. Note: You can restrict the balancer so that it only operates between specic start and end times. See Schedule the Balancing Window (page 528) for more information. The specication of the balancing window is relative to the local time zone of all individual mongos (page 1032) instances in the sharded cluster.
Chunk Size The default chunk size in MongoDB is 64 megabytes. When chunks grow beyond the specied chunk size (page 498) a mongos (page 1032) instance will split the chunk in half. This will eventually lead to migrations, when chunks become unevenly distributed among the cluster. The mongos (page 1032) instances will initiate a round of migrations to redistribute data in the cluster. Chunk size is arbitrary and must account for the following: 1. Small chunks lead to a more even distribution of data at the expense of more frequent migrations, which creates expense at the query routing (mongos (page 1032)) layer. 2. Large chunks lead to fewer migrations, which is more efcient both from the networking perspective and in terms internal overhead at the query routing layer. Large chunks produce these efciencies at the expense of a potentially more uneven distribution of data. For many deployments it makes sense to avoid frequent and potentially spurious migrations at the expense of a slightly less evenly distributed data set, but this value is congurable (page 523). Be aware of the following limitations when modifying chunk size: Automatic splitting only occurs when inserting documents or updating existing documents; if you lower the chunk size it may take time for all chunks to split to the new size. Splits cannot be undone: if you increase the chunk size, existing chunks must grow through insertion or updates until they reach the new size. Note: Chunk ranges are inclusive of the lower boundary and exclusive of the upper boundary.
Shard Size By default, MongoDB will attempt to ll all available disk space with data on every shard as the data set grows. Monitor disk utilization in addition to other performance metrics, to ensure that the cluster always has capacity to accommodate additional data. You can also congure a maximum size for any shard when you add the shard using the maxSize parameter of the addShard (page 846) command. This will prevent the balancer from migrating chunks to the shard when the value of mapped (page 894) exceeds the maxSize setting. See also:
498
Change the Maximum Storage Size for a Given Shard (page 526) and Monitoring for MongoDB (page 91). Chunk Migration MongoDB migrates chunks in a sharded cluster to distribute data evenly among shards. Migrations may be either: Manual. In these migrations you must specify the chunk that you want to migrate and the destination shard. Only migrate chunks manually after initiating sharding, to distribute data during bulk inserts, or if the cluster becomes uneven. See Migrating Chunks (page 524) for more details. Automatic. The balancer process handles most migrations when distribution of chunks between shards becomes uneven. See Migration Thresholds (page 497) for more details. All chunk migrations use the following procedure: 1. The balancer process sends the moveChunk (page 849) command to the source shard for the chunk. In this operation the balancer passes the name of the destination shard to the source shard. 2. The source initiates the move with an internal moveChunk (page 849) command with the destination shard. 3. The destination shard begins requesting documents in the chunk, and begins receiving these chunks. 4. After receiving the nal document in the chunk, the destination shard initiates a synchronization process to ensure that all changes to the documents in the chunk on the source shard during the migration process exist on the destination shard. When fully synchronized, the destination shard connects to the cong database and updates the chunk location in the cluster metadata. After completing this operation, once there are no open cursors on the chunk, the source shard starts deleting its copy of documents from the migrated chunk. If enabled, the _secondaryThrottle setting causes the balancer to wait for replication to secondaries. For more information, see Require Replication before Chunk Migration (Secondary Throttle) (page 527). Detect Connections to mongos Instances If your application must detect if the MongoDB instance its connected to is mongos (page 1032), use the isMaster (page 838) command. When a client connects to a mongos (page 1032), isMaster (page 838) returns a document with a msg eld that holds the string isdbgrid. For example:
{ "ismaster" : true, "msg" : "isdbgrid", "maxBsonObjectSize" : 16777216, "ok" : 1 }
If the application is instead connected to a mongod (page 1021), the returned document does not include the isdbgrid string.
40.5.3 Sharded Cluster Metadata

Sharded cluster metadata is contained in the Cong Database (page 545) and comprises information about the sharded clusters partitioned data sets. The cong database stores the relationship between chunks and where they reside within a sharded cluster. Without a cong database, the mongos (page 1032) instances would be unable to route queries or write operations within the cluster.
499
Cong Database The config database contains information about your sharding conguration and stores the information in a set of collections used by sharding. Important: Back up the config database before performing any maintenance on the cong server. To access the config database, issue the following command from the mongo (page 1036) shell:
use config
In general, you should never manipulate the content of the cong database directly. The config database contains the following collections: changelog (page 545) chunks (page 547) collections (page 548) databases (page 548) lockpings (page 548) locks (page 548) mongos (page 549) settings (page 549) shards (page 549) version (page 550) See Cong Database (page 545) for full documentation of these collections and their role in sharded clusters. Cong Servers Cong servers are special mongod (page 1021) instances that maintain the sharded cluster metadata in the cong database. A sharded cluster operates with a group of three cong servers that use a two-phase commit process that ensures immediate consistency and reliability. Cong servers do not run as replica sets. For testing purposes you may deploy a cluster with a single cong server, but this is not recommended for production. All cong servers must be available on initial setup of a sharded cluster. Each mongos (page 1032) instance must be able to write to the config.version collection. Warning: If your cluster has a single cong server, this mongod (page 1021) is a single point of failure. If the instance is inaccessible the cluster is not accessible. If you cannot recover the data on a cong server, the cluster will be inoperable. Always use three cong servers for production deployments.
Read and Write Operations on Cong Servers The load on conguration servers is small because each mongos (page 1032) instance maintains a cached copy of the conguration database. MongoDB only writes data to the cong server to: create splits in existing chunks, which happens as data in existing chunks exceeds the maximum chunk size. migrate a chunk between shards. 500 Chapter 40. Sharding Concepts
If one or two conguration instances become unavailable, the clusters metadata becomes read only. It is still possible to read and write data from the shards, but no chunk migrations or splits will occur until all three servers are accessible. At the same time, cong server data is only read in the following situations: A new mongos (page 1032) starts for the rst time, or an existing mongos (page 1032) restarts. After a chunk migration, the mongos (page 1032) instances update themselves with the new cluster metadata. If all three cong servers are inaccessible, you can continue to use the cluster as long as you dont restart the mongos (page 1032) instances until after cong servers are accessible again. If you restart the mongos (page 1032) instances and there are no accessible cong servers, the mongos (page 1032) would be unable to direct queries or write operations to the cluster. Because the conguration data is small relative to the amount of data stored in a cluster, the amount of activity is relatively low, and 100% up time is not required for a functioning sharded cluster. As a result, backing up the cong servers is not difcult. Backups of cong servers are critical as clusters become totally inoperable when you lose all conguration instances and data. Precautions to ensure that the cong servers remain available and intact are critical. Note: Conguration servers store metadata for a single sharded cluster. You must have a separate conguration server or servers for each cluster you administer.
501
502
CHAPTER 41
Administration
The following document provides a list of tutorials for administering sharded clusters.
41.1 Sharded Cluster Administration

The following tutorials provide instructions for administering sharded clusters. For a higher-level overview, see Sharding (page 483).
41.1.1 Deploy Sharded Clusters

Deploy a Sharded Cluster This document presents an ordered sequence of the tasks required to set up a sharded cluster. Before deploying a sharded cluster for the rst time, consider the Sharded Cluster Overview (page 485) and Sharded Cluster Architectures (page 487) documents. To set up a sharded cluster, complete the following sequence of tasks in the order dened below: 1. Start the Cong Server Database Instances (page 503) 2. Start the mongos Instances (page 504) 3. Add Shards to the Cluster (page 504) 4. Enable Sharding for a Database (page 505) 5. Enable Sharding for a Collection (page 506) Warning: Sharding and localhost Addresses If you use either localhost or 127.0.0.1 as the hostname portion of any host identier, for example as the host argument to addShard (page 846) or the value to the --configdb (page 1034) run time option, then you must use localhost or 127.0.0.1 for all host settings for any MongoDB instances in the cluster. If you mix localhost addresses and remote host address, MongoDB will error.
Start the Cong Server Database Instances
The cong server processes are mongod (page 1021) instances that store the clusters metadata. You designate a mongod (page 1021) as a cong server using the --configsvr (page 1030) option. Each cong server stores a complete copy of the clusters metadata. 503
In production deployments, you must deploy exactly three cong server instances, each running on different servers to assure good uptime and data safety. In test environments, you can run all three instances on a single server. Cong server instances receive relatively little trafc and demand only a small portion of system resources. Therefore, you can run an instance on a system that runs other cluster components. 1. Create data directories for each of the three cong server instances. By default, a cong server stores its data les in the /data/congdb directory. You can choose a different location. To create a data directory, issue a command similar to the following:
mkdir /data/configdb
2. Start the three cong server instances. Start each by issuing a command using the following syntax:
mongod --configsvr --dbpath <path> --port <port>
The default port for cong servers is 27019. You can specify a different port. The following example starts a cong server using the default port and default data directory:
mongod --configsvr --dbpath /data/configdb --port 27019
For additional command options, see mongod (page 1021) or Conguration File Options (page 1078). Note: All cong servers must be running and available when you rst initiate a sharded cluster.
Start the mongos Instances
The mongos (page 1032) instances are lightweight and do not require data directories. You can run a mongos (page 1032) instance on a system that runs other cluster components, such as on an application server or a server running a mongod (page 1021) process. By default, a mongos (page 1032) instance runs on port 27017. When you start the mongos (page 1032) instance, specify the hostnames of the three cong servers, either in the conguration le or as command line parameters. For operational exibility, use DNS names for the cong servers rather than explicit IP addresses. If youre not using resolvable hostname, you cannot change the cong server names or IP addresses without a restarting every mongos (page 1032) and mongod (page 1021) instance. To start a mongos (page 1032) instance, issue a command using the following syntax:
mongos --configdb <config server hostnames>
For example, to start a mongos (page 1032) that connects to cong server instance running on the following hosts and on the default ports: cfg0.example.net cfg1.example.net cfg2.example.net You would issue the following command:
mongos --configdb cfg0.example.net:27019,cfg1.example.net:27019,cfg2.example.net:27019
Add Shards to the Cluster
A shard can be a standalone mongod (page 1021) or a replica set. In a production environment, each shard should be a replica set.
504
Chapter 41. Administration
1. From a mongo (page 1036) shell, connect to the mongos (page 1032) instance. Issue a command using the following syntax:
mongo --host <hostname of machine running mongos> --port <port mongos listens on>
For example, if a mongos (page 1032) is accessible at mongos0.example.net on port 27017, issue the following command:
mongo --host mongos0.example.net --port 27017
2. Add each shard to the cluster using the sh.addShard() (page 995) method, as shown in the examples below. Issue sh.addShard() (page 995) separately for each shard. If the shard is a replica set, specify the name of the replica set and specify a member of the set. In production deployments, all shards should be replica sets. Optional You can instead use the addShard (page 846) database command, which lets you specify a name and maximum size for the shard. If you do not specify these, MongoDB automatically assigns a name and maximum size. To use the database command, see addShard (page 846). The following are examples of adding a shard with sh.addShard() (page 995): To add a shard for a replica set named rs1 with a member running on port 27017 on mongodb0.example.net, issue the following command:
sh.addShard( "rs1/mongodb0.example.net:27017" )
Changed in version 2.0.3. For MongoDB versions prior to 2.0.3, you must specify all members of the replica set. For example:
sh.addShard( "rs1/mongodb0.example.net:27017,mongodb1.example.net:27017,mongodb2.example.net
To add a shard for a standalone mongod (page 1021) on port 27017 of mongodb0.example.net, issue the following command:
sh.addShard( "mongodb0.example.net:27017" )
Note: It might take some time for chunks to migrate to the new shard.
Enable Sharding for a Database
Before you can shard a collection, you must enable sharding for the collections database. Enabling sharding for a database does not redistribute data but make it possible to shard the collections in that database. Once you enable sharding for a database, MongoDB assigns a primary shard for that database where MongoDB stores all data before sharding begins. 1. From a mongo (page 1036) shell, connect to the mongos (page 1032) instance. Issue a command using the following syntax:
2. Issue the sh.enableSharding() (page 998) method, specifying the name of the database for which to enable sharding. Use the following syntax:
41.1. Sharded Cluster Administration
505
sh.enableSharding("<database>")
Optionally, you can enable sharding for a database using the enableSharding (page 848) command, which uses the following syntax:
db.runCommand( { enableSharding: <database> } )
Enable Sharding for a Collection
You enable sharding on a per-collection basis. 1. Determine what you will use for the shard key. Your selection of the shard key affects the efciency of sharding. See the selection considerations listed in the Select a Shard Key (page 507). 2. If the collection already contains data you must create an index on the shard key using ensureIndex() (page 921). If the collection is empty then MongoDB will create the index as part of the sh.shardCollection() (page 1000) step. 3. Enable sharding for a collection by issuing the sh.shardCollection() (page 1000) method in the mongo (page 1036) shell. The method uses the following syntax:
sh.shardCollection("<database>.<collection>", shard-key-pattern)
Replace the <database>.<collection> string with the full namespace of your database, which consists of the name of your database, a dot (e.g. .), and the full name of the collection. The shard-key-pattern represents your shard key, which you specify in the same form as you would an index (page 921) key pattern. Example The following sequence of commands shards four collections:
sh.shardCollection("records.people", { "zipcode": 1, "name": 1 } ) sh.shardCollection("people.addresses", { "state": 1, "_id": 1 } ) sh.shardCollection("assets.chairs", { "type": 1, "_id": 1 } ) db.alerts.ensureIndex( { _id : "hashed" } ) sh.shardCollection("events.alerts", { "_id": "hashed" } )
In order, these operations shard: (a) The people collection in the records database using the shard key { "zipcode": 1 }. 1, "name":
This shard key distributes documents by the value of the zipcode eld. If a number of documents have the same value for this eld, then that chunk will be splitable (page 493) by the values of the name eld. (b) The addresses collection in the people database using the shard key { "state": 1 }. 1, "_id":
This shard key distributes documents by the value of the state eld. If a number of documents have the same value for this eld, then that chunk will be splitable (page 493) by the values of the _id eld. (c) The chairs collection in the assets database using the shard key { "type": }. 1, "_id": 1
This shard key distributes documents by the value of the type eld. If a number of documents have the same value for this eld, then that chunk will be splitable (page 493) by the values of the _id eld.
506
(d) The alerts collection in the events database using the shard key { "_id": New in version 2.4.
"hashed" }.
This shard key distributes documents by a hash of the value of the _id eld. MongoDB computes the hash of the _id eld for the hashed index (page 350), which should provide an even distribution of documents across a cluster. Select a Shard Key This document gives guidelines for selecting a shard key. Choosing the correct shard key can have a great impact on the performance, capability, and functioning of your database and cluster. Appropriate shard key choice depends on the schema of your data and the way that your application queries and writes data to the database. Use the following guidelines when creating a shard key.
Create a Shard Key that is Easily Divisible
An easily divisible shard key makes it easy for MongoDB to distribute content among the shards. Shard keys that have a limited number of possible values can result in chunks that are unsplitable. See the Cardinality (page 493) section for more information.
Create a Shard Key that has High Randomness
A shard key with high randomness prevents any single shard from becoming a bottleneck and will distribute write operations among the cluster. Conversely, a shard keys that has a high correlation with insert time is a poor choice. For more information, see the Write Scaling (page 494).
Create a Shard Key that Targets a Single Shard
A shard key that targets a single shard makes it possible for the mongos (page 1032) program to return most query operations directly from a single specic mongod (page 1021) instance. Your shard key should be the primary eld used by your queries. Fields with a high degree of randomness are poor choices for this reason. For examples, see Query Isolation (page 494).
Create a Special Purpose or Compound Key
The challenge when selecting a shard key is that there is not always an obvious choice. Often, an existing eld in your collection may not be the optimal key. In those situations, computing a special purpose shard key into an additional eld or using a compound shard key may help produce one that is more ideal. Hashed Sharding New in version 2.4. Hashed shard keys (page 496) use a hashed index (page 350) of a single eld as the shard key to partition data across your sharded cluster.
507
Procedure
To shard a collection using a hashed shard key, issue an operation in the mongo (page 1036) shell that resembles the following:
sh.shardCollection( "records.active", { a: "hashed" } )
This operation shards the active collection in the records database, using a hash of the a eld as the shard key.
Additional Considerations
The eld you choose as your hashed shard key should have a good cardinality, or large number of different values. Hashed keys work well with elds that increase monotonically like ObjectId values or timestamps. If you shard an empty collection using a hashed shard key, MongoDB will automatically create and migrate chunks so that each shard has two chunks. You can control how many chunks MongoDB will create with the numInitialChunks parameter to shardCollection (page 851). See Create a Hashed Index (page 350) for limitations on hashed indexes. Warning: hashed indexes truncate oating point numbers to 64-bit integers before hashing. For example, a hashed index would store the same value for a eld that held a value of 2.3, 2.2 and 2.9. To prevent collisions, do not use a hashed index for oating point numbers that cannot be consistently converted to 64-bit integers (and then back to oating point.) hashed indexes do not support oating point values larger than 253 . Warning: Hashed shard keys are only supported by the MongoDB 2.4 and greater versions of the mongos (page 1032) program. After sharding a collection with a hashed shard key, you must use the MongoDB 2.4 or greater mongos (page 1032) instances in your sharded cluster.
Enable Authentication in a Sharded Cluster New in version 2.0: Support for authentication with sharded clusters. To control access to a sharded cluster, create key les and then set the keyFile (page 1081) option on all components of the sharded cluster, including all mongos (page 1032) instances, all cong server mongod (page 1021) instances, and all shard mongod (page 1021) instances. The content of the key le is arbitrary but must be the same on all cluster members. Note: For an overview of authentication, see Access Control (page 135). For an overview of security, see Security (page 129).
Procedure
To enable authentication, do the following: 1. Generate a key le to store authentication information, as described in the Generate a Key File (page 148) section. 2. On each component in the sharded cluster, enable authentication by doing one of the following: In the conguration le, set the keyFile (page 1081) option to the key les path and then start the component, as in the following example:
508
When starting the component, set --keyFile (page 1034) option, which is an option for both mongos (page 1032) instances and mongod (page 1021) instances. Set the --keyFile (page 1034) to the key les path. Note: The keyFile (page 1081) setting implies auth (page 1081), which means in most cases you do not need to set auth (page 1081) explicitly. 3. Add the rst administrative user and then add subsequent users. See Create an Administrator with Superuser Credentials (page 146). View Cluster Conguration
List Databases with Sharding Enabled
To list the databases that have sharding enabled, query the databases collection in the Cong Database (page 545). A database has sharding enabled if the value of the partitioned eld is true. Connect to a mongos (page 1032) instance with a mongo (page 1036) shell, and run the following operation to get a full list of databases with sharding enabled:
use config db.databases.find( { "partitioned": true } )
Example You can use the following sequence of commands when to return a list of all databases in the cluster:
use config db.databases.find()
If this returns the following result set:

{ "_id" : "admin", "partitioned" : false, "primary" : "config" } { "_id" : "animals", "partitioned" : true, "primary" : "m0.example.net:30001" } { "_id" : "farms", "partitioned" : false, "primary" : "m1.example2.net:27017" }
Then sharding is only enabled for the animals database.
List Shards
To list the current set of congured shards, use the listShards (page 849) command, as follows:
use admin db.runCommand( { listShards : 1 } )
View Cluster Details
To view cluster details, issue db.printShardingStatus() (page 983) or sh.status() (page 1002). Both methods return the same output. Example In the following example output from sh.status() (page 1002) 41.1. Sharded Cluster Administration 509
sharding version displays the version number of the shard metadata. shards displays a list of the mongod (page 1021) instances used as shards in the cluster. databases displays all databases in the cluster, including database that do not have sharding enabled. The chunks information for the foo database displays how many chunks are on each shard and displays the range of each chunk.
--- Sharding Status --sharding version: { "_id" : 1, "version" : 3 } shards: { "_id" : "shard0000", "host" : "m0.example.net:30001" } { "_id" : "shard0001", "host" : "m3.example2.net:50000" } databases: { "_id" : "admin", "partitioned" : false, "primary" : "config" } { "_id" : "animals", "partitioned" : true, "primary" : "shard0000" } foo.big chunks: shard0001 1 shard0000 6 { "a" : { $minKey : 1 } } -->> { "a" : "elephant" } on : shard0001 Timestamp(2000, 1) jum { "a" : "elephant" } -->> { "a" : "giraffe" } on : shard0000 Timestamp(1000, 1) jumbo { "a" : "giraffe" } -->> { "a" : "hippopotamus" } on : shard0000 Timestamp(2000, 2) jumbo { "a" : "hippopotamus" } -->> { "a" : "lion" } on : shard0000 Timestamp(2000, 3) jumbo { "a" : "lion" } -->> { "a" : "rhinoceros" } on : shard0000 Timestamp(1000, 3) jumbo { "a" : "rhinoceros" } -->> { "a" : "springbok" } on : shard0000 Timestamp(1000, 4) { "a" : "springbok" } -->> { "a" : { $maxKey : 1 } } on : shard0000 Timestamp(1000, 5) foo.large chunks: shard0001 1 shard0000 5 { "a" : { $minKey : 1 } } -->> { "a" : "hen" } on : shard0001 Timestamp(2000, 0) { "a" : "hen" } -->> { "a" : "horse" } on : shard0000 Timestamp(1000, 1) jumbo { "a" : "horse" } -->> { "a" : "owl" } on : shard0000 Timestamp(1000, 2) jumbo { "a" : "owl" } -->> { "a" : "rooster" } on : shard0000 Timestamp(1000, 3) jumbo { "a" : "rooster" } -->> { "a" : "sheep" } on : shard0000 Timestamp(1000, 4) { "a" : "sheep" } -->> { "a" : { $maxKey : 1 } } on : shard0000 Timestamp(1000, 5) { "_id" : "test", "partitioned" : false, "primary" : "shard0000" }
Add Shards to a Cluster You add shards to a sharded cluster after you create the cluster or anytime that you need to add capacity to the cluster. If you have not created a sharded cluster, see Deploy a Sharded Cluster (page 503). When adding a shard to a cluster, you should always ensure that the cluster has enough capacity to support the migration without affecting legitimate production trafc. In production environments, all shards should be replica sets.
Add a Shard to a Cluster
You interact with a sharded cluster by connecting to a mongos (page 1032) instance. 1. From a mongo (page 1036) shell, connect to the mongos (page 1032) instance. Issue a command using the following syntax:
510
For example, if a mongos (page 1032) is accessible at mongos0.example.net on port 27017, issue the following command:
mongo --host mongos0.example.net --port 27017
2. Add each shard to the cluster using the sh.addShard() (page 995) method, as shown in the examples below. Issue sh.addShard() (page 995) separately for each shard. If the shard is a replica set, specify the name of the replica set and specify a member of the set. In production deployments, all shards should be replica sets. Optional You can instead use the addShard (page 846) database command, which lets you specify a name and maximum size for the shard. If you do not specify these, MongoDB automatically assigns a name and maximum size. To use the database command, see addShard (page 846). The following are examples of adding a shard with sh.addShard() (page 995): To add a shard for a replica set named rs1 with a member running on port 27017 on mongodb0.example.net, issue the following command:
sh.addShard( "rs1/mongodb0.example.net:27017" )
Changed in version 2.0.3. For MongoDB versions prior to 2.0.3, you must specify all members of the replica set. For example:
sh.addShard( "rs1/mongodb0.example.net:27017,mongodb1.example.net:27017,mongodb2.example.net
To add a shard for a standalone mongod (page 1021) on port 27017 of mongodb0.example.net, issue the following command:
sh.addShard( "mongodb0.example.net:27017" )
Note: It might take some time for chunks to migrate to the new shard.
Convert a Replica Set to a Replicated Sharded Cluster

Overview
Following this tutorial, you will convert a single 3-member replica set to a cluster that consists of 2 shards. Each shard will consist of an independent 3-member replica set. The tutorial uses a test environment running on a local system UNIX-like system. You should feel encouraged to follow along at home. If you need to perform this process in a production environment, notes throughout the document indicate procedural differences. The procedure, from a high level, is as follows: 1. Create or select a 3-member replica set and insert some data into a collection. 2. Start the cong databases and create a cluster with a single shard. 3. Create a second replica set with three new mongod (page 1021) instances. 4. Add the second replica set as a shard in the cluster. 5. Enable sharding on the desired collection or collections.
511
Process
Install MongoDB according to the instructions in the MongoDB Installation Tutorial (page 3). Deploy a Replica Set with Test Data If have an existing MongoDB replica set deployment, you can omit the this step and continue from Deploy Sharding Infrastructure (page 513). Use the following sequence of steps to congure and deploy a replica set and to insert test data. 1. Create the following directories for the rst replica set instance, named firstset: http://docs.mongodb.org/manual/data/example/firstset1 http://docs.mongodb.org/manual/data/example/firstset2 http://docs.mongodb.org/manual/data/example/firstset3 To create directories, issue the following command:
mkdir -p /data/example/firstset1 /data/example/firstset2 /data/example/firstset3
2. In a separate terminal window or GNU Screen window, start three mongod (page 1021) instances by running each of the following commands:
mongod --dbpath /data/example/firstset1 --port 10001 --replSet firstset --oplogSize 700 --rest mongod --dbpath /data/example/firstset2 --port 10002 --replSet firstset --oplogSize 700 --rest mongod --dbpath /data/example/firstset3 --port 10003 --replSet firstset --oplogSize 700 --rest
Note: The --oplogSize 700 (page 1028) option restricts the size of the operation log (i.e. oplog) for each mongod (page 1021) instance to 700MB. Without the --oplogSize (page 1028) option, each mongod (page 1021) reserves approximately 5% of the free disk space on the volume. By limiting the size of the oplog, each instance starts more quickly. Omit this setting in production environments. 3. In a mongo (page 1036) shell session in a new terminal, connect to the mongodb instance on port 10001 by running the following command. If you are in a production environment, rst read the note below.
mongo localhost:10001/admin
Note: Above and hereafter, if you are running in a production environment or are testing this process with mongod (page 1021) instances on multiple systems, replace localhost with a resolvable domain, hostname, or the IP address of your system. 4. In the mongo (page 1036) shell, initialize the rst replica set by issuing the following command:
db.runCommand({"replSetInitiate" : {"_id" : "firstset", "members" : [{"_id" : 1, "host" {"_id" : 2, "host" {"_id" : 3, "host" ]}}) { "info" : "Config now saved locally. Should come online in about "ok" : 1 } : "localhost:10001"}, : "localhost:10002"}, : "localhost:10003"}
a minute.",
5. In the mongo (page 1036) shell, create and populate a new collection by issuing the following sequence of JavaScript operations:
512
use test switched to db test people = ["Marc", "Bill", "George", "Eliot", "Matt", "Trey", "Tracy", "Greg", "Steve", "Kristina for(var i=0; i<1000000; i++){ name = people[Math.floor(Math.random()*people.length)]; user_id = i; boolean = [true, false][Math.floor(Math.random()*2)]; added_at = new Date(); number = Math.floor(Math.random()*10001); db.test_collection.save({"name":name, "user_id":user_id, "boolean": }
The above operations add one million documents to the collection test_collection. This can take several minutes, depending on your system. The script adds the documents in the following form:
{ "_id" : ObjectId("4ed5420b8fc1dd1df5886f70"), "name" : "Greg", "user_id" : 4, "boolean" : true, "ad
Deploy Sharding Infrastructure This procedure creates the three cong databases that store the clusters metadata. Note: For development and testing environments, a single cong database is sufcient. In production environments, use three cong databases. Because cong instances store only the metadata for the sharded cluster, they have minimal resource requirements. 1. Create the following data directories for three cong database instances: http://docs.mongodb.org/manual/data/example/config1 http://docs.mongodb.org/manual/data/example/config2 http://docs.mongodb.org/manual/data/example/config3 Issue the following command at the system prompt:
mkdir -p /data/example/config1 /data/example/config2 /data/example/config3
2. In a separate terminal window or GNU Screen window, start the cong databases by running the following commands:
mongod --configsvr --dbpath /data/example/config1 --port 20001 mongod --configsvr --dbpath /data/example/config2 --port 20002 mongod --configsvr --dbpath /data/example/config3 --port 20003
3. In a separate terminal window or GNU Screen window, start mongos (page 1032) instance by running the following command:
mongos --configdb localhost:20001,localhost:20002,localhost:20003 --port 27017 --chunkSize 1
Note: If you are using the collection created earlier or are just experimenting with sharding, you can use a small --chunkSize (page 1034) (1MB works well.) The default chunkSize (page 1090) of 64MB means that your cluster must have 64MB of data before the MongoDBs automatic sharding begins working. In production environments, do not use a small shard size. The configdb (page 1089) options specify the conguration databases (e.g. localhost:20001, localhost:20002, and localhost:2003). The mongos (page 1032) instance runs on the default MongoDB port (i.e. 27017), while the databases themselves are running on ports in the 30001 series. In the this example, you may omit the --port 27017 (page 1032) option, as 27017 is the default port. 41.1. Sharded Cluster Administration 513
4. Add the rst shard in mongos (page 1032). In a new terminal window or GNU Screen session, add the rst shard, according to the following procedure: (a) Connect to the mongos (page 1032) with the following command:
(b) Add the rst shard to the cluster by issuing the addShard (page 846) command:
db.runCommand( { addShard : "firstset/localhost:10001,localhost:10002,localhost:10003" } )
(c) Observe the following message, which denotes success:

{ "shardAdded" : "firstset", "ok" : 1 }
Deploy a Second Replica Set This procedure deploys a second replica set. This closely mirrors the process used to establish the rst replica set above, omitting the test data. 1. Create the following data directories for the members of the second replica set, named secondset: http://docs.mongodb.org/manual/data/example/secondset1 http://docs.mongodb.org/manual/data/example/secondset2 http://docs.mongodb.org/manual/data/example/secondset3 2. In three new terminal windows, start three instances of mongod (page 1021) with the following commands:
mongod --dbpath /data/example/secondset1 --port 10004 --replSet secondset --oplogSize 700 --rest mongod --dbpath /data/example/secondset2 --port 10005 --replSet secondset --oplogSize 700 --rest mongod --dbpath /data/example/secondset3 --port 10006 --replSet secondset --oplogSize 700 --rest
Note: As above, the second replica set uses the smaller oplogSize (page 1088) conguration. Omit this setting in production environments. 3. In the mongo (page 1036) shell, connect to one mongodb instance by issuing the following command:
4. In the mongo (page 1036) shell, initialize the second replica set by issuing the following command:
db.runCommand({"replSetInitiate" : {"_id" : "secondset", "members" : [{"_id" : 1, "host" : "localhost:10004"}, {"_id" : 2, "host" : "localhost:10005"}, {"_id" : 3, "host" : "localhost:10006"} ]}}) { "info" : "Config now saved locally. "ok" : 1 } Should come online in about a minute.",
5. Add the second replica set to the cluster. Connect to the mongos (page 1032) instance created in the previous procedure and issue the following sequence of commands:
use admin db.runCommand( { addShard : "secondset/localhost:10004,localhost:10005,localhost:10006" } )
This command returns the following success message:
514
{ "shardAdded" : "secondset", "ok" : 1 }
6. Verify that both shards are properly congured by running the listShards (page 849) command. View this and example output below:
db.runCommand({listShards:1}) { "shards" : [ { "_id" : "firstset", "host" : "firstset/localhost:10001,localhost:10003,localhost:10002" }, { "_id" : "secondset", "host" : "secondset/localhost:10004,localhost:10006,localhost:10005" } ], "ok" : 1 }
Enable Sharding MongoDB must have sharding enabled on both the database and collection levels. Enabling Sharding on the Database Level Issue the enableSharding (page 848) command. The following example enables sharding on the test database:
db.runCommand( { enableSharding : "test" } ) { "ok" : 1 }
Create an Index on the Shard Key MongoDB uses the shard key to distribute documents between shards. Once selected, you cannot change the shard key. Good shard keys: have values that are evenly distributed among all documents, group documents that are often accessed at the same time into contiguous chunks, and allow for effective distribution of activity among shards. Typically shard keys are compound, comprising of some sort of hash and some sort of other primary key. Selecting a shard key depends on your data set, application architecture, and usage pattern, and is beyond the scope of this document. For the purposes of this example, we will shard the number key. This typically would not be a good shard key for production deployments. Create the index with the following procedure:
use test db.test_collection.ensureIndex({number:1})
See also: The Shard Key Overview (page 485) and Shard Key (page 493) sections. Shard the Collection Issue the following command:
use admin db.runCommand( { shardCollection : "test.test_collection", key : {"number":1} }) { "collectionsharded" : "test.test_collection", "ok" : 1 }
515
The collection test_collection is now sharded! Over the next few minutes the Balancer begins to redistribute chunks of documents. You can conrm this activity by switching to the test database and running db.stats() (page 986) or db.printShardingStatus() (page 983). As clients insert additional documents into this collection, mongos (page 1032) distributes the documents evenly between the shards. In the mongo (page 1036) shell, issue the following commands to return statics against each cluster:
use test db.stats() db.printShardingStatus()
Example output of the db.stats() (page 986) command:

{ "raw" : { "firstset/localhost:10001,localhost:10003,localhost:10002" : { "db" : "test", "collections" : 3, "objects" : 973887, "avgObjSize" : 100.33173458522396, "dataSize" : 97711772, "storageSize" : 141258752, "numExtents" : 15, "indexes" : 2, "indexSize" : 56978544, "fileSize" : 1006632960, "nsSizeMB" : 16, "ok" : 1 }, "secondset/localhost:10004,localhost:10006,localhost:10005" : { "db" : "test", "collections" : 3, "objects" : 26125, "avgObjSize" : 100.33286124401914, "dataSize" : 2621196, "storageSize" : 11194368, "numExtents" : 8, "indexes" : 2, "indexSize" : 2093056, "fileSize" : 201326592, "nsSizeMB" : 16, "ok" : 1 } }, "objects" : 1000012, "avgObjSize" : 100.33176401883178, "dataSize" : 100332968, "storageSize" : 152453120, "numExtents" : 23, "indexes" : 4, "indexSize" : 59071600, "fileSize" : 1207959552, "ok" : 1 }
Example output of the db.printShardingStatus() (page 983) command:
516
--- Sharding Status --sharding version: { "_id" : 1, "version" : 3 } shards: { "_id" : "firstset", "host" : "firstset/localhost:10001,localhost:10003,localhost:10002" } { "_id" : "secondset", "host" : "secondset/localhost:10004,localhost:10006,localhost:10005" databases: { "_id" : "admin", "partitioned" : false, "primary" : "config" } { "_id" : "test", "partitioned" : true, "primary" : "firstset" } test.test_collection chunks: secondset 5 firstset 186 [...]
In a few moments you can run these commands for a second time to demonstrate that chunks are migrating from firstset to secondset. When this procedure is complete, you will have converted a replica set into a cluster where each shard is itself a replica set.
41.1.2 Sharded Cluster Maintenance and Administration

Administer and Manage Shard Tags In a sharded cluster, you can use tags to associate specic ranges of a shard key with a specic shard or subset of shards.
Tag a Shard
Associate tags with a particular shard using the sh.addShardTag() (page 996) method when connected to a mongos (page 1032) instance. A single shard may have multiple tags, and multiple shards may also have the same tag. Example The following example adds the tag NYC to two shards, and the tags SFO and NRT to a third shard:
sh.addShardTag("shard0000", sh.addShardTag("shard0001", sh.addShardTag("shard0002", sh.addShardTag("shard0002", "NYC") "NYC") "SFO") "NRT")
You may remove tags from a particular shard using the sh.removeShardTag() (page 1000) method when connected to a mongos (page 1032) instance, as in the following example, which removes the NRT tag from a shard:
sh.removeShardTag("shard0002", "NRT")
Tag a Shard Key Range
To assign a tag to a range of shard keys use the sh.addTagRange() method when connected to a mongos (page 1032) instance. Any given shard key range may only have one assigned tag. You cannot overlap dened ranges, or tag the same range more than once.
517
Example Given a collection named users in the records database, sharded by the zipcode eld. The following operations assign: two ranges of zip codes in Manhattan and Brooklyn the NYC tag one range of zip codes in San Francisco the SFO tag
sh.addTagRange("records.users", { zipcode: "10001" }, { zipcode: "10281" }, "NYC") sh.addTagRange("records.users", { zipcode: "11201" }, { zipcode: "11240" }, "NYC") sh.addTagRange("records.users", { zipcode: "94102" }, { zipcode: "94135" }, "SFO")
Note: Shard ranges are always inclusive of the lower value and exclusive of the upper boundary.
Remove a Tag From a Shard Key Range
The mongod (page 1021) does not provide a helper for removing a tag range. You may delete tag assignment from a shard key range by removing the corresponding document from the tags (page 550) collection of the config database. Each document in the tags (page 550) holds the namespace of the sharded collection and a minimum shard key value. Example The following example removes the NYC tag assignment for the range of zip codes within Manhattan:
use config db.tags.remove({ _id: { ns: "records.users", min: { zipcode: "10001" }}, tag: "NYC" })
View Existing Shard Tags
The output from sh.status() (page 1002) lists tags associated with a shard, if any, for each shard. A shards tags exist in the shards document in the shards (page 549) collection of the config database. To return all shards with a specic tag, use a sequence of operations that resemble the following, which will return only those shards tagged with NYC:
use config db.shards.find({ tags: "NYC" })
You can nd tag ranges for all namespaces in the tags (page 550) collection of the config database. The output of sh.status() (page 1002) displays all tag ranges. To return all shard key ranges tagged with NYC, use the following sequence of operations:
use config db.tags.find({ tags: "NYC" })
Deploy Three Cong Servers for Production Deployments This procedure converts a test deployment with only one cong server (page 500) to a production deployment with three cong servers.
518
For redundancy, all production sharded clusters (page 485) should deploy three cong servers on three different machines. Use a single cong server only for testing deployments, never for production deployments. When you shift to production, upgrade immediately to three cong servers. To convert a test deployment with one cong server to a production deployment with three cong servers: 1. Shut down all existing MongoDB processes in the cluster. This includes: all mongod (page 1021) instances or replica sets that provide your shards. all mongos (page 1032) instances in your cluster. 2. Copy the entire dbpath (page 1081) le system tree from the existing cong server to the two machines that will provide the additional cong servers. These commands, issued on the system with the existing Cong Database (page 545), mongo-config0.example.net may resemble the following:
rsync -az /data/configdb mongo-config1.example.net:/data/configdb rsync -az /data/configdb mongo-config2.example.net:/data/configdb
3. Start all three cong servers, using the same invocation that you used for the single cong server.
mongod --configsvr
4. Restart all shard mongod (page 1021) and mongos (page 1032) processes. Migrate Cong Servers with the Same Hostname This procedure migrates a cong server (page 500) in a sharded cluster (page 485) to a new system that uses the same hostname. 1. Shut down the cong server that you are moving. This will render all cong data for your cluster read only (page 500). 2. Change the DNS entry that points to the system that provided the old cong server, so that the same hostname points to the new system. How you do this depends on how you organize your DNS and hostname resolution services. 3. Move the entire dbpath (page 1081) le system tree from the old cong server to the new cong server. This command, issued on the old cong server system, may resemble the following:
rsync -az /data/configdb mongo-config0.example.net:/data/configdb
4. Start the cong instance on the new system. The default invocation is:
mongod --configsvr
When you start the third cong server, your cluster will become writable and it will be able to create new splits and migrate chunks as needed. Migrate Cong Servers with Different Hostnames This procedure migrates a cong server (page 500) in a sharded cluster (page 485) to a new server that uses a different hostname. Use this procedure only if the cong server will not be accessible via the same hostname. If possible, avoid changing the hostname so that you can instead use the procedure to migrate a cong server and use the same hostname (page 519). 1. Disable the cluster balancer process temporarily. See Disable the Balancer (page 529) for more information.
519
2. Shut down the cong server (page 500) you are moving. This will render all cong data for your cluster read only:
rsync -az /data/configdb mongodb.config2.example.net:/data/configdb
3. Start the cong instance on the new system. The default invocation is:
mongod --configsvr
4. Shut down all existing MongoDB processes. This includes: all mongod (page 1021) instances or replica sets that provide your shards. the mongod (page 1021) instances that provide your existing cong databases (page 545). all mongos (page 1032) instances in your cluster. 5. Restart all mongod (page 1021) processes that provide the shard servers. 6. Update the --configdb (page 1034) parameter (or configdb (page 1089)) for all mongos (page 1032) instances and restart all mongos (page 1032) instances. 7. Re-enable the balancer to allow the cluster to resume normal balancing operations. See the Disable the Balancer (page 529) section for more information on managing the balancer process. Replace a Cong Server This procedure replaces an inoperable cong server (page 500) in a sharded cluster (page 485). Use this procedure only to replace a cong server that has become inoperable (e.g. hardware failure). This process assumes that the hostname of the instance will not change. If you must change the hostname of the instance, use the procedure to migrate a cong server and use a new hostname (page 519). 1. Disable the cluster balancer process temporarily. See Disable the Balancer (page 529) for more information. 2. Provision a new system, with the same hostname as the previous host. You will have to ensure that the new system has the same IP address and hostname as the system its replacing or you will need to modify the DNS records and wait for them to propagate. 3. Shut down one (and only one) of the existing cong servers. Copy all this hosts dbpath (page 1081) le system tree from the current system to the system that will provide the new cong server. This command, issued on the system with the data les, may resemble the following:
rsync -az /data/configdb mongodb.config2.example.net:/data/configdb
4. Restart the cong server process that you used in the previous step to copy the data les to the new cong server instance. 5. Start the new cong server instance. The default invocation is:
mongod --configsvr
6. Re-enable the balancer to allow the cluster to resume normal balancing operations. See the Disable the Balancer (page 529) section for more information on managing the balancer process. Note: In the course of this procedure never remove a cong server from the configdb (page 1089) parameter on any of the mongos (page 1032) instances. If you need to change the name of a cong server, always make sure that all mongos (page 1032) instances have three cong servers specied in the configdb (page 1089) setting at all times.
520
Backup Cluster Metadata This procedure shuts down the mongod (page 1021) instance of a cong server (page 500) in order to create a backup of a sharded clusters (page 485) metadata. The clusters cong servers store all of the clusters metadata, most importantly the mapping from chunks to shards. When you perform this procedure, the cluster remains operational 1 . 1. Disable the cluster balancer process temporarily. See Disable the Balancer (page 529) for more information. 2. Shut down one of the cong databases. 3. Create a full copy of the data les (i.e. the path specied by the dbpath (page 1081) option for the cong instance.) 4. Restart the original conguration server. 5. Re-enable the balancer to allow the cluster to resume normal balancing operations. See the Disable the Balancer (page 529) section for more information on managing the balancer process. See also: Backup Strategies for MongoDB Systems (page 41). Manage Chunks in a Sharded Cluster This page describes various operations on chunks in sharded clusters. MongoDB automates most chunk management operations. However, these chunk management operations are accessible to administrators for use in some situations, typically surrounding initial setup, deployment, and data ingestion.
Split Chunks
Normally, MongoDB splits a chunk following inserts when a chunk exceeds the chunk size (page 498). The balancer may migrate recently split chunks to a new shard immediately if mongos (page 1032) predicts future insertions will benet from the move. MongoDB treats all chunks the same, whether split manually or automatically by the system. Warning: You cannot merge or combine chunks once you have split them. You may want to split chunks manually if: you have a large amount of data in your cluster and very few chunks, as is the case after deploying a cluster using existing data. you expect to add a large amount of data that would initially reside in a single chunk or shard. Example You plan to insert a large amount of data with shard key values between 300 and 400, but all values of your shard keys are between 250 and 500 are in a single chunk.
1 While one of the three cong servers is unavailable, the cluster cannot split any chunks nor can it migrate chunks between shards. Your application will be able to write data to the cluster. See Cong Servers (page 500) for more information.
521
Warning: Be careful when splitting data in a sharded collection to create new chunks. When you shard a collection that has existing data, MongoDB automatically creates chunks to evenly distribute the collection. To split data effectively in a sharded cluster you must consider the number of documents in a chunk and the average document size to create a uniform chunk size. When chunks have irregular sizes, shards may have an equal number of chunks but have very different data sizes. Avoid creating splits that lead to a collection with differently sized chunks. Use sh.status() (page 1002) to determine the current chunks ranges across the cluster. To split chunks manually, use the split (page 853) command with operators: middle and find. The equivalent shell helpers are sh.splitAt() (page 1001) or sh.splitFind() (page 1001). Example The following command will split the chunk that contains the value of 63109 for the zipcode eld in the people collection of the records database:
sh.splitFind( "records.people", { "zipcode": 63109 } )
sh.splitFind() (page 1001) will split the chunk that contains the rst document returned that matches this query into two equally sized chunks. You must specify the full namespace (i.e. <database>.<collection>) of the sharded collection to sh.splitFind() (page 1001). The query in sh.splitFind() (page 1001) need not contain the shard key, though it almost always makes sense to query for the shard key in this case, and including the shard key will expedite the operation. Use sh.splitAt() (page 1001) to split a chunk in two using the queried document as the partition point:
sh.splitAt( "records.people", { "zipcode": 63109 } )
However, the location of the document that this query nds with respect to the other documents in the chunk does not affect how the chunk splits.
Create Chunks (Pre-Splitting)
Pre-splitting the chunk ranges in an empty sharded collection, allows clients to insert data into an already-partitioned collection. In most situations a sharded cluster will create and distribute chunks automatically without user intervention. However, in a limited number of use proles, MongoDB cannot create enough chunks or distribute data fast enough to support required throughput. For example, if: you must partition an existing data collection that resides on a single shard. you must ingest a large volume of data into a cluster that isnt balanced, or where the ingestion of data will lead to an imbalance of data. This can arise in an initial data loading, or in a case where you must insert a large volume of data into a single chunk, as is the case when you must insert at the beginning or end of the chunk range, as is the case for monotonically increasing or decreasing shard keys. Preemptively splitting chunks increases cluster throughput for these operations, by reducing the overhead of migrating chunks that hold data during the write operation. MongoDB only creates splits after an insert operation and can migrate only a single chunk at a time. Chunk migrations are resource intensive and further complicated by large write volume to the migrating chunk. Warning: You can only pre-split an empty collection. When you enable sharding for a collection that contains data MongoDB automatically creates splits. Subsequent attempts to create splits manually, can lead to unpredictable chunk ranges and sizes as well as inefcient or ineffective balancing behavior.
522
To create and migrate chunks manually, use the following procedure: 1. Split empty chunks in your collection by manually performing split (page 853) command on chunks. Example To create chunks for documents in the myapp.users collection, using the email eld as the shard key, use the following operation in the mongo (page 1036) shell:
for ( var x=97; x<97+26; x++ ){ for( var y=97; y<97+26; y+=6 ) { var prefix = String.fromCharCode(x) + String.fromCharCode(y); db.runCommand( { split : "myapp.users" , middle : { email : prefix } } ); } }
This assumes a collection size of 100 million documents. 2. Migrate chunks manually using the moveChunk (page 849) command: Example To migrate all of the manually created user proles evenly, putting each prex chunk on the next shard from the other, run the following commands in the mongo shell:
var shServer = [ "sh0.example.net", "sh1.example.net", "sh2.example.net", "sh3.example.net", for ( var x=97; x<97+26; x++ ){ for( var y=97; y<97+26; y+=6 ) { var prefix = String.fromCharCode(x) + String.fromCharCode(y); db.adminCommand({moveChunk : "myapp.users", find : {email : prefix}, to : shServer[(y-97 } }
You can also let the balancer automatically distribute the new chunks. For an introduction to balancing, see Shard Balancing (page 486). For lower level information on balancing, see Cluster Balancer (page 497).
Modify Chunk Size
When you initialize a sharded cluster, 2 the default chunk size is 64 megabytes. This default chunk size works well for most deployments; however, if you notice that automatic migrations are incurring a level of I/O that your hardware cannot handle, you may want to reduce the chunk size. For the automatic splits and migrations, a small chunk size leads to more rapid and frequent migrations. to modify the chunk size, use the following procedure: 1. connect to any mongos (page 1032) in the cluster using the mongo (page 1036) shell. 2. issue the following command to switch to the Cong Database (page 545):
use config
3. Issue the following save() (page 946) operation:

db.settings.save( { _id:"chunksize", value: <size> } )
Where the value of <size> reects the new chunk size in megabytes. Here, youre essentially writing a document whose values store the global chunk size conguration value.
2
The rst mongos (page 1032) that connects to a set of cong servers initializes the sharded cluster.
523
Note: The chunkSize (page 1090) and --chunkSize (page 1034) options, passed at runtime to the mongos (page 1032) do not affect the chunk size after you have initialized the cluster. 1 To eliminate confusion you should always set chunk size using the above procedure and never use the runtime options. Modifying the chunk size has several limitations: Automatic splitting only occurs when inserting documents or updating existing documents. If you lower the chunk size it may take time for all chunks to split to the new size. Splits cannot be undone. If you increase the chunk size, existing chunks must grow through insertion or updates until they reach the new size.
Migrate Chunks
In most circumstances, you should let the automatic balancer migrate chunks between shards. However, you may want to migrate chunks manually in a few cases: If you create chunks by pre-splitting the data in your collection, you will have to migrate chunks manually to distribute chunks evenly across the shards. Use pre-splitting in limited situations, to support bulk data ingestion. If the balancer in an active cluster cannot distribute chunks within the balancing window, then you will have to migrate chunks manually. For more information on how chunks move between shards, see Cluster Balancer (page 497), in particular the section Chunk Migration (page 499). To migrate chunks, use the moveChunk (page 849) command. Note: To return a list of shards, use the listShards (page 849) command. Specify shard names using the addShard (page 846) command using the name argument. If you do not specify a name in the addShard (page 846) command, MongoDB will assign a name automatically. The following example assumes that the eld username is the shard key for a collection named users in the myapp database, and that the value smith exists within the chunk you want to migrate. To move this chunk, you would issue the following command from a mongo (page 1036) shell connected to any mongos (page 1032) instance.
db.adminCommand( { moveChunk : "myapp.users", find : {username : "smith"}, to : "mongodb-shard3.example.net" } )
This command moves the chunk that includes the shard key value smith to the shard named mongodb-shard3.example.net. The command will block until the migration is complete. See Create Chunks (Pre-Splitting) (page 522) for an introduction to pre-splitting. New in version 2.2: moveChunk (page 849) command has the: _secondaryThrottle parameter. When set to true, MongoDB ensures that changes to shards as part of chunk migrations replicate to secondaries throughout the migration operation. For more information, see Require Replication before Chunk Migration (Secondary Throttle) (page 527). Changed in version 2.4: In 2.4, _secondaryThrottle is true by default.
524
Warning: The moveChunk (page 849) command may produce the following error message:
The collections metadata lock is already taken.
These errors occur when clients have too many open cursors that access the chunk you are migrating. You can either wait until the cursors complete their operation or close the cursors manually.
Strategies for Bulk Inserts in Sharded Clusters
Large bulk insert operations, including initial data ingestion or routine data import, can have a signicant impact on a sharded cluster. For bulk insert operations, consider the following strategies: If the collection does not have data, then there is only one chunk, which must reside on a single shard. MongoDB must receive data, create splits, and distribute chunks to the available shards. To avoid this performance cost, you can pre-split the collection, as described in Create Chunks (Pre-Splitting) (page 522). You can parallelize import processes by sending insert operations to more than one mongos (page 1032) instance. If the collection is empty, pre-split rst, as described in Create Chunks (Pre-Splitting) (page 522). If your shard key increases monotonically during an insert then all the inserts will go to the last chunk in the collection, which will always end up on a single shard. Therefore, the insert capacity of the cluster will never exceed the insert capacity of a single shard. If your insert volume is never larger than what a single shard can process, then there is no problem; however, if the insert volume exceeds that range, and you cannot avoid a monotonically increasing shard key, then consider the following modications to your application: Reverse all the bits of the shard key to preserve the information while avoiding the correlation of insertion order and increasing sequence of values. Swap the rst and last 16-bit words to shufe the inserts. Example The following example, in C++, swaps the leading and trailing 16-bit word of BSON ObjectIds generated so that they are no longer monotonically increasing.
using namespace mongo; OID make_an_id() { OID x = OID::gen(); const unsigned char *p = x.getData(); swap( (unsigned short&) p[0], (unsigned short&) p[10] ); return x; } void foo() { // create an object BSONObj o = BSON( "_id" << make_an_id() << "x" << 3 << "name" << "jane" ); // now we might insert o into a sharded collection... }
For information on choosing a shard key, see Shard Keys (page 485) and see Shard Key Internals (page 493) (in particular, Operations and Reliability (page 495) and Choosing a Shard Key (page 495)).
525
Congure Behavior of Balancer Process in Sharded Clusters The balancer is a process that runs on one of the mongos (page 1032) instances in a cluster and ensures that chunks are evenly distributed throughout a sharded cluster. In most deployments, the default balancer conguration is sufcient for normal operation. However, administrators might need to modify balancer behavior depending on application or operational requirements. If you encounter a situation where you need to modify the behavior of the balancer, use the procedures described in this document. For conceptual information about the balancer, see Shard Balancing (page 486) and Cluster Balancer (page 497).
Schedule a Window of Time for Balancing to Occur
You can schedule a window of time during which the balancer can migrate chunks, as described in the following procedures: Schedule the Balancing Window (page 528) Remove a Balancing Window Schedule (page 528). The mongos (page 1032) instances user their own local timezones to when respecting balancer window.
Congure Default Chunk Size
The default chunk size for a sharded cluster is 64 megabytes. In most situations, the default size is appropriate for splitting and migrating chunks. For information on how chunk size affects deployments, see details, see Chunk Size (page 498). Changing the default chunk size affects chunks that are processes during migrations and auto-splits but does not retroactively affect all chunks. To congure default chunk size, see Modify Chunk Size (page 523).
Change the Maximum Storage Size for a Given Shard
The maxSize eld in the shards (page 549) collection in the cong database (page 545) sets the maximum size for a shard, allowing you to control whether the balancer will migrate chunks to a shard. If dataSize (page 878) is above a shards maxSize, the balancer will not move chunks to the shard. Also, the balancer will not move chunks off an overloaded shard. This must happen manually. The maxSize value only affects the balancers selection of destination shards. By default, maxSize is not specied, allowing shards to consume the total amount of available space on their machines if necessary. You can set maxSize both when adding a shard and once a shard is running. To set maxSize when adding a shard, set the addShard (page 846) commands maxSize parameter to the maximum size in megabytes. For example, the following command run in the mongo (page 1036) shell adds a shard with a maximum size of 125 megabytes:
db.runCommand( { addshard : "example.net:34008", maxSize : 125 } )
To set maxSize on an existing shard, insert or update the maxSize eld in the shards (page 549) collection in the cong database (page 545). Set the maxSize in megabytes. Example Assume you have the following shard without a maxSize eld:
526
{ "_id" : "shard0000", "host" : "example.net:34001" }
Run the following sequence of commands in the mongo (page 1036) shell to insert a maxSize of 125 megabytes:
use config db.shards.update( { _id : "shard0000" }, { $set : { maxSize : 125 } } )
To later increase the maxSize setting to 250 megabytes, run the following:
use config db.shards.update( { _id : "shard0000" }, { $set : { maxSize : 250 } } )
Require Replication before Chunk Migration (Secondary Throttle)
New in version 2.2.1: _secondaryThrottle became an option to the balancer and to moveChunk (page 849) in 2.2.1. _secondaryThrottle makes it possible to require the balancer wait for replication to secondaries during migrations. Changed in version 2.4: _secondaryThrottle became the default mode for all balancer and moveChunk (page 849) operations. See also: Secondary Throttle in the v2.2 Manual for more information on conguring _secondaryThrottle. Manage Sharded Cluster Balancer This page describes provides common administrative procedures related to balancing. For an introduction to balancing, see Shard Balancing (page 486). For lower level information on balancing, see Cluster Balancer (page 497). See also: Congure Behavior of Balancer Process in Sharded Clusters (page 526)
Check the Balancer Lock
To see if the balancer process is active in your cluster, do the following: 1. Connect to any mongos (page 1032) in the cluster using the mongo (page 1036) shell. 2. Issue the following command to switch to the Cong Database (page 545):
use config
3. Use the following query to return the balancer lock:

db.locks.find( { _id : "balancer" } ).pretty()
When this command returns, you will see output like the following:
{ "_id" "process" "state" "ts" "when" "who" "why" : : : : : : : "balancer", "mongos0.example.net:1292810611:1804289383", 2, ObjectId("4d0f872630c42d1978be8a2e"), "Mon Dec 20 2010 11:41:10 GMT-0500 (EST)", "mongos0.example.net:1292810611:1804289383:Balancer:846930886", "doing balance round" }
527
This output conrms that: The balancer originates from the mongos (page 1032) running on the system with the hostname mongos0.example.net. The value in the state eld indicates that a mongos (page 1032) has the lock. For version 2.0 and later, the value of an active lock is 2; for earlier versions the value is 1. Optional You can also use the following shell helper, which returns a boolean to report if the balancer is active:
sh.getBalancerState()
Schedule the Balancing Window
In some situations, particularly when your data set grows slowly and a migration can impact performance, its useful to be able to ensure that the balancer is active only at certain times. Use the following procedure to specify a window during which the balancer will be able to migrate chunks: 1. Connect to any mongos (page 1032) in the cluster using the mongo (page 1036) shell. 2. Issue the following command to switch to the Cong Database (page 545):
use config
3. Use an operation modeled on the following example update() (page 948) operation to modify the balancers window:
db.settings.update({ _id : "balancer" }, { $set : { activeWindow : { start : "<start-time>", sto
Replace <start-time> and <end-time> with time values using two digit hour and minute values (e.g HH:MM) that describe the beginning and end boundaries of the balancing window. These times will be evaluated relative to the time zone of each individual mongos (page 1032) instance in the sharded cluster. If your mongos (page 1032) instances are physically located in different time zones, use a common time zone (e.g. GMT) to ensure that the balancer window is interpreted correctly. For instance, running the following will force the balancer to run between 11PM and 6AM local time only:
db.settings.update({ _id : "balancer" }, { $set : { activeWindow : { start : "23:00", stop : "6:
Note: The balancer window must be sufcient to complete the migration of all data inserted during the day. As data insert rates can change based on activity and usage patterns, it is important to ensure that the balancing window you select will be sufcient to support the needs of your deployment.
Remove a Balancing Window Schedule
If you have set the balancing window (page 528) and wish to remove the schedule so that the balancer is always running, issue the following sequence of operations:
use config db.settings.update({ _id : "balancer" }, { $unset : { activeWindow : true } })
528
Disable the Balancer
By default the balancer may run at any time and only moves chunks as needed. To disable the balancer for a short period of time and prevent all migration, use the following procedure: 1. Connect to any mongos (page 1032) in the cluster using the mongo (page 1036) shell. 2. Issue the following operation to disable the balancer:
sh.stopBalancer()
3. Later, issue the following operation to enable the balancer:

sh.startBalancer()
Note: If a migration is in progress, the system will complete the in-progress migration. After disabling, you can use the following operation in the mongo (page 1036) shell to determine if there are no migrations in progress:
use config while( db.locks.findOne({_id: "balancer"}).state ) { print("waiting..."); sleep(1000); }
The above process and the sh.setBalancerState() (page 1000), sh.startBalancer() (page 1002), and sh.stopBalancer() (page 1004) helpers provide wrappers on the following process, which may be useful if you need to run this operation from a driver that does not have helper functions: 1. Connect to any mongos (page 1032) in the cluster using the mongo (page 1036) shell. 2. Issue the following command to switch to the Cong Database (page 545):
use config
3. Issue the following update to disable the balancer:

db.settings.update( { _id: "balancer" }, { $set : { stopped: true } } , true );
4. To enable the balancer again, alter the value of stopped as follows:

db.settings.update( { _id: "balancer" }, { $set : { stopped: false } } , true );
Disable Balancing During Backups
If MongoDB migrates a chunk during a backup (page 41), you can end with an inconsistent snapshot of your sharded cluster. Never run a backup while the balancer is active. To ensure that the balancer is inactive during your backup operation: Set the balancing window (page 528) so that the balancer is inactive during the backup. Ensure that the backup can complete while you have the balancer disabled. manually disable the balancer (page 529) for the duration of the backup procedure. Conrm that the balancer is not active using the sh.getBalancerState() (page 998) method before starting a backup operation. When the backup procedure is complete you can reactivate the balancer process.
529
Remove Shards from an Existing Sharded Cluster To remove a shard you must ensure the shards data is migrated to the remaining shards in the cluster. This procedure describes how to safely migrate data and how to remove a shard. This procedure describes how to safely remove a single shard. Do not use this procedure to migrate an entire cluster to new hardware. To migrate an entire shard to new hardware, migrate individual shards as if they were independent replica sets. To remove a shard, rst connect to one of the clusters mongos (page 1032) instances using mongo (page 1036) shell. Then follow the ordered sequence of tasks on this page: 1. Ensure the Balancer Process is Active (page 530) 2. Determine the Name of the Shard to Remove (page 530) 3. Remove Chunks from the Shard (page 530) 4. Check the Status of the Migration (page 531) 5. Move Unsharded Data (page 531) 6. Finalize the Migration (page 531)
Ensure the Balancer Process is Active
To successfully migrate data from a shard, the balancer process must be active. Check the balancer state using the sh.getBalancerState() (page 998) helper in the mongo (page 1036) shell. For more information, see the section on balancer operations (page 529).
Determine the Name of the Shard to Remove
To determine the name of the shard, connect to a mongos (page 1032) instance with the mongo (page 1036) shell and either: Use the listShards (page 849) command, as in the following:
db.adminCommand( { listShards: 1 } )
Run either the sh.status() (page 1002) or the db.printShardingStatus() (page 983) method. The shards._id eld lists the name of each shard.
Remove Chunks from the Shard
Run the removeShard (page 850) command. This begins draining chunks from the shard you are removing to other shards in the cluster. For example, for a shard named mongodb0, run:
db.runCommand( { removeShard: "mongodb0" } )
This operation returns immediately, with the following response:

{ msg : "draining started successfully" , state: "started" , shard :"mongodb0" , ok : 1 }
Depending on your network capacity and the amount of data, this operation can take from a few minutes to several days to complete.
530
Check the Status of the Migration
To check the progress of the migration at any stage in the process, run removeShard (page 850). For example, for a shard named mongodb0, run:
The command returns output similar to the following:
{ msg: "draining ongoing" , state: "ongoing" , remaining: { chunks: NumberLong(42), dbs : NumberLong(
In the output, the remaining document displays the remaining number of chunks that MongoDB must migrate to other shards and the number of MongoDB databases that have primary status on this shard. Continue checking the status of the removeShard command until the number of chunks remaining is 0. Then proceed to the next step.
Move Unsharded Data
If the shard is the primary shard for one or more databases in the cluster, then the shard will have unsharded data. If the shard is not the primary shard for any databases, skip to the next task, Finalize the Migration (page 531). In a cluster, a database with unsharded collections stores those collections only on a single shard. That shard becomes the primary shard for that database. (Different databases in a cluster can have different primary shards.) Warning: Do not perform this procedure until you have nished draining the shard. 1. To determine if the shard you are removing is the primary shard for any of the clusters databases, issue one of the following methods: sh.status() (page 1002) db.printShardingStatus() (page 983) In the resulting document, the databases eld lists each database and its primary shard. For example, the following database eld shows that the products database uses mongodb0 as the primary shard:
{ "_id" : "products", "partitioned" : true, "primary" : "mongodb0" }
2. To move a database to another shard, use the movePrimary (page 850) command. For example, to migrate all remaining unsharded data from mongodb0 to mongodb1, issue the following command:
db.runCommand( { movePrimary: "products", to: "mongodb1" })
This command does not return until MongoDB completes moving all data, which may take a long time. The response from this command will resemble the following:
{ "primary" : "mongodb1", "ok" : 1 }
Finalize the Migration
To clean up all metadata information and nalize the removal, run removeShard (page 850) again. For example, for a shard named mongodb0, run:
A success message appears at completion:
531
{ msg: "remove shard completed successfully" , stage: "completed", host: "mongodb0", ok : 1 }
Once the value of the stage eld is completed, you may safely stop the processes comprising the mongodb0 shard. See also: Backup and Restore Sharded Clusters (page 54).
41.1.3 Manage Data in Sharded Clusters

Tag Aware Sharding For sharded clusters, MongoDB makes it possible to associate specic ranges of a shard key with a specic shard or subset of shards. This association dictates the policy of the cluster balancer process as it balances the chunks around the cluster. This capability enables the following deployment patterns: isolating a specic subset of data on specic set of shards. controlling the balancing policy so that, in a geographically distributed cluster, the most relevant portions of the data set reside on the shards with the greatest proximity to the application servers. This document describes the behavior, operation, and use of tag aware sharding in MongoDB deployments. Note: Shard key range tags are entirely distinct from replica set member tags (page 404). Hash-based sharding does not support tag-aware sharding.
Behavior and Operations
Tags in a sharded cluster are pieces of metadata that dictate the policy and behavior of the cluster balancer. Using tags, you may associate individual shards in a cluster with one or more tags. Then, you can assign this tag string to a range of shard key values for a sharded collection. When migrating a chunk, the balancer will select a destination shard based on the congured tag ranges. The balancer migrates chunks in tagged ranges to shards with those tags, if tagged shards are not balanced.
3
Note: Because a single chunk may span different tagged shard key ranges, the balancer may migrate chunks to tagged shards that contain values that exceed the upper bound of the selected tag range. Example Given a sharded collection with two congured tag ranges, such that: Shard key values between 100 and 200 have tags to direct corresponding chunks to shards tagged NYC. Shard Key values between 200 and 300 have tags to direct corresponding chunks to shards tagged SFO. In this cluster, the balancer will migrate a chunk with shard key values ranging between 150 and 220 to a shard tagged NYC, since 150 is closer to 200 than 300. After conguring tags on the shards and ranges of the shard key, the cluster may take some time to reach the proper distribution of data, depending on the division of chunks (i.e. splits) and the current distribution of data in the cluster. Once congured, the balancer will respect tag ranges during future balancing rounds (page 497).
3 To migrate chunks in a tagged environment, the balancer selects a target shard with a tag range that has an upper bound that is greater than the migrating chunks lower bound. If a shard with a matching tagged range exists, the balancer will migrate the chunk to that shard.
532
See also: Administer and Manage Shard Tags (page 517) Enforce Unique Keys for Sharded Collections
Overview
The unique (page 921) constraint on indexes ensures that only one document can have a value for a eld in a collection. For sharded collections these unique indexes cannot enforce uniqueness (page 1106) because insert and indexing operations are local to each shard. 4 If your need to ensure that a eld is always unique in all collections in a sharded environment, there are two options: 1. Enforce uniqueness of the shard key (page 485). MongoDB can enforce uniqueness for the shard key. For compound shard keys, MongoDB will enforce uniqueness on the entire key combination, and not for a specic component of the shard key. You cannot specify a unique constraint on a hashed index (page 335). 2. Use a secondary collection to enforce uniqueness. Create a minimal collection that only contains the unique eld and a reference to a document in the main collection. If you always insert into a secondary collection before inserting to the main collection, MongoDB will produce an error if you attempt to use a duplicate key. Note: If you have a small data set, you may not need to shard this collection and you can create multiple unique indexes. Otherwise you can shard on a single unique key. Always use the default acknowledged (page 399) write concern (page 398) in conjunction with a recent MongoDB driver (page 1185).
Unique Constraints on the Shard Key
Process To shard a collection using the unique constraint, specify the shardCollection (page 851) command in the following form:
db.runCommand( { shardCollection : "test.users" , key : { email : 1 } , unique : true } );
Remember that the _id eld index is always unique. By default, MongoDB inserts an ObjectId into the _id eld. However, you can manually insert your own value into the _id eld and use this as the shard key. To use the _id eld as the shard key, use the following operation:
db.runCommand( { shardCollection : "test.users" } )
Warning: In any sharded collection where you are not sharding by the _id eld, you must ensure uniqueness of the _id eld. The best way to ensure _id is always unique is to use ObjectId, or another universally unique identier (UUID.)
4 If you specify a unique index on a sharded collection, MongoDB will be able to enforce uniqueness only among the documents located on a single shard at the time of creation.
533
Limitations You can only enforce uniqueness on one single eld in the collection using this method. If you use a compound shard key, you can only enforce uniqueness on the combination of component keys in the shard key. In most cases, the best shard keys are compound keys that include elements that permit write scaling (page 494) and query isolation (page 494), as well as high cardinality (page 493). These ideal shard keys are not often the same keys that require uniqueness and requires a different approach.
Unique Constraints on Arbitrary Fields
If you cannot use a unique eld as the shard key or if you need to enforce uniqueness over multiple elds, you must create another collection to act as a proxy collection. This collection must contain both a reference to the original document (i.e. its ObjectId) and the unique key. If you must shard this proxy collection, then shard on the unique key using the above procedure (page 533); otherwise, you can simply create multiple unique indexes on the collection. Process Consider the following for the proxy collection:
{ "_id" : ObjectId("...") "email" ": "..." }
The _id eld holds the ObjectId of the document it reects, and the email eld is the eld on which you want to ensure uniqueness. To shard this collection, use the following operation using the email eld as the shard key:
db.runCommand( { shardCollection : "records.proxy" , key : { email : 1 } , unique : true } );
If you do not need to shard the proxy collection, use the following command to create a unique index on the email eld:
db.proxy.ensureIndex( { "email" : 1 }, {unique : true} )
You may create multiple unique indexes on this collection if you do not plan to shard the proxy collection. To insert documents, use the following procedure in the JavaScript shell (page 1036):
use records; var primary_id = ObjectId(); db.proxy.insert({ "_id" : primary_id "email" : "example@example.net" }) // if: the above operation returns successfully, // then continue: db.information.insert({ "_id" : primary_id "email": "example@example.net"
534
// additional information... })
You must insert a document into the proxy collection rst. If this operation succeeds, the email eld is unique, and you may continue by inserting the actual document into the information collection. See also: The full documentation of: (page 851). Considerations Your application must catch errors when inserting documents into the proxy collection and must enforce consistency between the two collections. If the proxy collection requires sharding, you must shard on the single eld on which you want to enforce uniqueness. To enforce uniqueness on more than one eld using sharded proxy collections, you must have one proxy collection for every eld for which to enforce uniqueness. If you create multiple unique indexes on a single proxy collection, you will not be able to shard proxy collections. Shard GridFS Data Store When sharding a GridFS store, consider the following:
files Collection
db.collection.ensureIndex() (page 921) and shardCollection
Most deployments will not need to shard the files collection. The files collection is typically small, and only contains metadata. None of the required keys for GridFS lend themselves to an even distribution in a sharded situation. If you must shard the files collection, use the _id eld possibly in combination with an application eld. Leaving files unsharded means that all the le metadata documents live on one shard. For production GridFS stores you must store the files collection on a replica set.
chunks Collection
To shard the chunks collection by { files_id :
1 , n :
1 }, issue commands similar to the following:
db.fs.chunks.ensureIndex( { files_id : 1 , n : 1 } ) db.runCommand( { shardCollection : "test.fs.chunks" , key : { files_id : 1 , n : 1 } } )
You may also want to shard using just the file_id eld, as in the following operation:
db.runCommand( { shardCollection : "test.fs.chunks" , key : { files_id : 1 } } )
Note: Changed in version 2.2. Before 2.2, you had to create an additional index on files_id to shard using only this eld. The default files_id value is an ObjectId, as a result the values of files_id are always ascending, and applications will insert all new GridFS data to a single chunk and shard. If your write load is too high for a single server to handle, consider a different shard key or use a different value for different value for _id in the files collection. 41.1. Sharded Cluster Administration 535
536
CHAPTER 42
Reference
Sharding Commands (page 537) Cong Database (page 545) mongos (page 1032) Cong Database (page 545)
42.1 Sharding Commands

42.1.1 JavaScript Methods
sh.addShard(host) Parameters host (string) Specify the hostname of a database instance or a replica set conguration. Use this method to add a database instance or replica set to a sharded cluster. This method must be run on a mongos (page 1032) instance. The host parameter can be in any of the following forms:
[hostname] [hostname]:[port] [set]/[hostname] [set]/[hostname],[hostname]:port
You can specify shards using the hostname, or a hostname and port combination if the shard is running on a non-standard port. Warning: Do not use localhost for the hostname unless your conguration server is also running on localhost. The optimal conguration is to deploy shards across replica sets. To add a shard on a replica set you must specify the name of the replica set and the hostname of at least one member of the replica set. You must specify at least one member of the set, but can specify all members in the set or another subset if desired. sh.addShard() (page 995) takes the following form: If you specify additional hostnames, all must be members of the same replica set.
sh.addShard("set-name/seed-hostname")
Example 537
sh.addShard("repl0/mongodb3.example.net:27327")
The sh.addShard() (page 995) method is a helper for the addShard (page 846) command. addShard (page 846) command has additional options which are not available with this helper. See also: addShard (page 846) Sharded Cluster Administration (page 503) Add Shards to a Cluster (page 510) Remove Shards from an Existing Sharded Cluster (page 530) sh.enableSharding(database) Parameters database (string) Specify a database name to shard.
The
Enables sharding on the specied database. This does not automatically shard any collections, but makes it possible to begin sharding collections using sh.shardCollection() (page 1000). See also: sh.shardCollection() (page 1000) sh.shardCollection(namespace, key, unique) Parameters namespace (string) The namespace of the collection to shard. key (document) A document containing a shard key that the sharding system uses to partition and distribute objects among the shards. unique (boolean) When true, the unique option ensures that the underlying index enforces a unique constraint. Hashed shard keys do not support unique constraints. Shards the named collection, according to the specied shard key. Specify shard keys in the form of a document. Shard keys may refer to a single document eld, or more typically several document elds to form a compound shard key. New in version 2.4: Use the form {field: may not be compound indexes. "hashed"} to create a hashed shard key. Hashed shard keys
Warning: MongoDB provides no method to deactivate sharding for a collection after calling shardCollection (page 851). Additionally, after shardCollection (page 851), you cannot change shard keys or modify the value of any eld used in your shard key index. See also: shardCollection (page 851) for additional options, Sharding (page 483), Sharded Cluster Overview (page 485) for an overview of sharding with MongoDB and Deploy a Sharded Cluster (page 503) for a tutorial. Also review Shard Keys (page 485) regarding choosing a shard key. sh.splitFind(namespace, query) Parameters namespace (string) Specify the namespace (i.e. <database>.<collection>) of the sharded collection that contains the chunk to split. 538 Chapter 42. Reference
query Specify a query to identify a document in a specic chunk. Typically specify the shard key for a document as the query. Splits the chunk containing the document specied by the query at its median point, creating two roughly equal chunks. Use sh.splitAt() (page 1001) to split a collection in a specic point. In most circumstances, you should leave chunk splitting to the automated processes. However, when initially deploying a sharded cluster it is necessary to perform some measure of pre-splitting using manual methods including sh.splitFind() (page 1001). sh.splitAt(namespace, query) Parameters namespace (string) Specify the namespace (i.e. <database>.<collection>) of the sharded collection that contains the chunk to split. query (document) Specify a query to identify a document in a specic chunk. Typically specify the shard key for a document as the query. Splits the chunk containing the document specied by the query as if that document were at the middle of the collection, even if the specied document is not the actual median of the collection. Use this command to manually split chunks unevenly. Use the sh.splitFind() (page 1001) function to split a chunk at the actual median. In most circumstances, you should leave chunk splitting to the automated processes within MongoDB. However, when initially deploying a sharded cluster it is necessary to perform some measure of pre-splitting using manual methods including sh.splitAt() (page 1001). sh.moveChunk(collection, query, destination) Parameters collection (string) Species the sharded collection containing the chunk to migrate. query (document) A document that species an equality match on the shard key, which selects the chunk to move. destination (string) Species the name of the shard that you wish to move the designated chunk to. Moves the chunk containing the document specied by the query to the shard described by destination. This method provides a wrapper around the moveChunk (page 849). In most circumstances, allow the balancer to automatically migrate chunks, and avoid calling sh.moveChunk() (page 999) directly. See also: moveChunk (page 849), sh.splitAt() (page 1001), sh.splitFind() (page 1001), Sharding (page 483), and chunk migration (page 499). sh.setBalancerState(state) Parameters state (boolean) true enables the balancer if disabled, and false disables the balancer. Enables or disables the balancer. Use sh.getBalancerState() (page 998) to determine if the balancer is currently enabled or disabled and sh.isBalancerRunning() (page 999) to check its current state. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997)
42.1. Sharding Commands
539
sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005) sh.isBalancerRunning() Returns boolean Returns true if the balancer process is currently running and migrating chunks and false if the balancer process is not running. Use sh.getBalancerState() (page 998) to determine if the balancer is enabled or disabled. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.setBalancerState() (page 1000) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005) sh.status() Prints a formatted report of the sharding conguration and the information regarding existing chunks in a sharded cluster. The default behavior suppresses the detailed chunk information if the total number of chunks is greater than or equal to 20. Parameters verbose (boolean) Optional. If true, the method displays details of the document distribution across chunks when you have 20 or more chunks. See also: db.printShardingStatus() (page 983) sh.addShardTag(shard, tag) New in version 2.2. Parameters shard (string) Species the name of the shard that you want to give a specic tag. tag (string) Species the name of the tag that you want to add to the shard. sh.addShardTag() (page 996) associates a shard with a tag or identier. MongoDB uses these identiers to direct chunks that fall within a tagged range to specic shards. sh.addTagRange() associates chunk ranges with tag ranges.
540
Always issue sh.addShardTag() (page 996) when connected to a mongos (page 1032) instance. Example The following example adds three tags, NYC, LAX, and NRT, to three shards:
sh.addShardTag("shard0000", "NYC") sh.addShardTag("shard0001", "LAX") sh.addShardTag("shard0002", "NRT")
See also: sh.addTagRange() and sh.removeShardTag() (page 1000) sh.addTagRange(namespace, minimum, maximum, tag) New in version 2.2. Parameters namespace (string) Species the namespace, in the form of <database>.<collection> of the sharded collection that you would like to tag. minimum (document) Species the minimum value of the shard key range to include in the tag. Specify the minimum value in the form of <fieldname>:<value>. This value must be of the same BSON type or types as the shard key. maximum (document) Species the maximum value of the shard key range to include in the tag. Specify the maximum value in the form of <fieldname>:<value>. This value must be of the same BSON type or types as the shard key. tag (string) Species the name of the tag to attach the range specied by the minimum and maximum arguments to. sh.addTagRange() attaches a range of values of the shard key to a shard tag created using the sh.addShardTag() (page 996) method. Use this operation to ensure that the documents that exist within the specied range exist on shards that have a matching tag. Always issue sh.addTagRange() when connected to a mongos (page 1032) instance. Note: If you add a tag range to a collection using sh.addTagRange(), and then later drop the collection or its database, MongoDB does not remove tag association. If you later create a new collection with the same name, the old tag association will apply to the new collection. Example Given a shard key of {STATE:1,ZIP:1}, create a tag range covering ZIP codes in New York State:
sh.addTagRange( "exampledb.collection", { STATE: "NY", ZIP: MinKey }, { STATE:"NY", ZIP: MaxKey }, "NY" )
See also: sh.addShardTag() (page 996), sh.removeShardTag() (page 1000)
541
sh.removeShardTag(shard, tag) New in version 2.2. Parameters shard (string) Species the name of the shard that you want to remove a tag from. tag (string) Species the name of the tag that you want to remove from the shard. Removes the association between a tag and a shard. Always issue sh.removeShardTag() (page 1000) when connected to a mongos (page 1032) instance. See also: sh.addShardTag() (page 996), sh.addTagRange() sh.help() Returns a basic help text for all sharding related shell functions.
42.1.2 Database Commands

The following database commands support sharded clusters. addShard Parameters hostname (string) a hostname or replica-set/hostname string. name (string) Optional. Unless specied, a name will be automatically provided to uniquely identify the shard. maxSize (integer) Optional, megabytes. Limits the maximum size of a shard. If maxSize is 0 then MongoDB will not limit the size of the shard. Use the addShard (page 846) command to add a database instance or replica set to a sharded cluster. You must run this command when connected a mongos (page 1032) instance. The command takes the following form:
{ addShard: "<hostname><:port>" }
Example
db.runCommand({addShard: "mongodb0.example.net:27027"})
Replace <hostname><:port> with the hostname and port of the database instance you want to add as a shard. Warning: Do not use localhost for the hostname unless your conguration server is also running on localhost. The optimal conguration is to deploy shards across replica sets. To add a shard on a replica set you must specify the name of the replica set and the hostname of at least one member of the replica set. You must specify at least one member of the set, but can specify all members in the set or another subset if desired. addShard (page 846) takes the following form:
{ addShard: "replica-set/hostname:port" }
542
Example
db.runCommand( { addShard: "repl0/mongodb3.example.net:27327"} )
If you specify additional hostnames, all must be members of the same replica set. Send this command to only one mongos (page 1032) instance, it will store shard conguration information in the cong database. Note: Specify a maxSize when you have machines with different disk capacities, or if you want to limit the amount of data on some shards. The maxSize constraint prevents the balancer from migrating chunks to the shard when the value of mem.mapped (page 894) exceeds the value of maxSize. See also: sh.addShard() (page 995) Sharded Cluster Administration (page 503) Add Shards to a Cluster (page 510) Remove Shards from an Existing Sharded Cluster (page 530) listShards Use the listShards (page 849) command to return a list of congured shards. The command takes the following form:
{ listShards: 1 }
enableSharding The enableSharding (page 848) command enables sharding on a per-database level. Use the following command form:
{ enableSharding: "<database name>" }
Once youve enabled sharding in a database, you can use the shardCollection (page 851) command to begin the process of distributing data among the shards. shardCollection The shardCollection (page 851) command marks a collection for sharding and will allow data to begin distributing among shards. You must run enableSharding (page 848) on a database before running the shardCollection (page 851) command.
{ shardCollection: "<database>.<collection>", key: <shardkey> }
This enables sharding for the collection specied by <collection> in the database named <database>, using the key <shardkey> to distribute documents among the shard. <shardkey> is a document and takes the same form as an index specication document (page 192). Parameters shardCollection (string) Specify the namespace of a collection to shard in the form <database>.<collection>. key (document) Specify the index specication to use as the shard key. The index must exist prior to the shardCollection (page 851) command unless the collection is empty. If the collection is empty, then MongoDB will create the index prior to sharding the collection.
543
New in version 2.4: The key may be in the form { field : use the specied eld as a hashed shard key (page 496) .
"hashed" } which will
unique (boolean) When true, the unique option ensures that the underlying index enforces a unique constraint. Hashed shard keys do not support unique constraints. numInitialChunks (integer) New in version 2.4. Specify the number of chunks to create upon sharding the collection. The collection will then be pre-split and balanced across the specied number of chunks. You can specify no more than 8192 chunks using numInitialChunks. Choosing the right shard key to effectively distribute load among your shards requires some planning. Also review Shard Keys (page 485) regarding choosing a shard key. Warning: MongoDB provides no method to deactivate sharding for a collection after calling shardCollection (page 851). Additionally, after shardCollection (page 851), you cannot change shard keys or modify the value of any eld used in your shard key index. See also: Sharding (page 483), Sharded Cluster Overview (page 485), and Deploy a Sharded Cluster (page 503). shardingState shardingState (page 852) is an admin command that reports if mongod (page 1021) is a member of a sharded cluster. shardingState (page 852) has the following prototype form:
{ shardingState: 1 }
For shardingState (page 852) to detect that a mongod (page 1021) is a member of a sharded cluster, the mongod (page 1021) must satisfy the following conditions: 1.the mongod (page 1021) is a primary member of a replica set, and 2.the mongod (page 1021) instance is a member of a sharded cluster. If shardingState (page 852) detects that a mongod (page 1021) is a member of a sharded cluster, shardingState (page 852) returns a document that resembles the following prototype:
{ "enabled" : true, "configServer" : "<configdb-string>", "shardName" : "<string>", "shardHost" : "string:", "versions" : { "<database>.<collection>" : Timestamp(<...>), "<database>.<collection>" : Timestamp(<...>) }, "ok" : 1 }
Otherwise, shardingState (page 852) will return the following document:

{ "note" : "from execCommand", "ok" : 0, "errmsg" : "not master" }
The response from shardingState (page 852) when used with a cong server is:
{ "enabled": false, "ok": 1 }
Note: mongos (page 1032) instances do not provide the shardingState (page 852).
544
Warning: This command obtains a write lock on the affected database and will block other operations until it has completed; however, the operation is typically short lived. removeShard Starts the process of removing a shard from a cluster. This is a multi-stage process. Begin by issuing the following command:
{ removeShard : "[shardName]" }
The balancer will then migrate chunks from the shard specied by [shardName]. This process happens slowly to avoid placing undue load on the overall cluster. The command returns immediately, with the following message:
{ msg : "draining started successfully" , state: "started" , shard: "shardName" , ok : 1 }
If you run the command again, youll see the following progress output:
{ msg: "draining ongoing" , state: "ongoing" , remaining: { chunks: 23 , dbs: 1 }, ok: 1 }
The remaining document species how many chunks and databases remain on the shard. db.printShardingStatus() (page 983) to list the databases that you must move from the shard.
Use
Each database in a sharded cluster has a primary shard. If the shard you want to remove is also the primary of one of the clusters databases, then you must manually move the database to a new shard. This can be only after the shard is empty. See the movePrimary (page 850) command for details. After removing all chunks and databases from the shard, you may issue the command again, to return:
{ msg: "remove shard completed successfully", stage: "completed", host: "shardName", ok : 1 }
42.2 Cong Database

The config database supports sharded cluster operation. See the Sharding (page 483) section of this manual for full documentation of sharded clusters. Warning: Consider the schema of the config database internal and may change between releases of MongoDB. The config database is not a dependable API, and users should not write data to the config database in the course of normal operation or maintenance. Modication of the config database on a functioning system may lead to instability or inconsistent data sets. To access the config database, connect to a mongos (page 1032) instance in a sharded cluster, and use the following helper:
use config
You can return a list of the collections, with the following helper:
show collections
42.2.1 Collections
config
42.2. Cong Database
545
config.changelog
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The changelog (page 545) collection stores a document for each change to the metadata of a sharded collection. Example The following example displays a single record of a chunk split from a changelog (page 545) collection:
{ "_id" : "<hostname>-<timestamp>-<increment>", "server" : "<hostname><:port>", "clientAddr" : "127.0.0.1:63381", "time" : ISODate("2012-12-11T14:09:21.039Z"), "what" : "split", "ns" : "<database>.<collection>", "details" : { "before" : { "min" : { "<database>" : { $minKey : 1 } }, "max" : { "<database>" : { $maxKey : 1 } }, "lastmod" : Timestamp(1000, 0), "lastmodEpoch" : ObjectId("000000000000000000000000") }, "left" : { "min" : { "<database>" : { $minKey : 1 } }, "max" : { "<database>" : "<value>" }, "lastmod" : Timestamp(1000, 1), "lastmodEpoch" : ObjectId(<...>) }, "right" : { "min" : { "<database>" : "<value>" }, "max" : { "<database>" : { $maxKey : 1 } }, "lastmod" : Timestamp(1000, 2), "lastmodEpoch" : ObjectId("<...>") } } }
Each document in the changelog (page 545) collection contains the following elds: config.changelog._id 546 Chapter 42. Reference
The value of changelog._id is: <hostname>-<timestamp>-<increment>. config.changelog.server The hostname of the server that holds this data. config.changelog.clientAddr A string that holds the address of the client, a mongos (page 1032) instance that initiates this change. config.changelog.time A ISODate timestamp that reects when the change occurred. config.changelog.what Reects the type of change recorded. Possible values are: dropCollection dropCollection.start dropDatabase dropDatabase.start moveChunk.start moveChunk.commit split multi-split config.changelog.ns Namespace where the change occurred. config.changelog.details A document that contains additional details regarding the change. (page 547) document depends on the type of change. config.chunks The structure of the details
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The chunks (page 547) collection stores a document for each chunk in the cluster. Consider the following example of a document for a chunk named records.pets-animal_\"cat\":
{ "_id" : "mydb.foo-a_\"cat\"", "lastmod" : Timestamp(1000, 3), "lastmodEpoch" : ObjectId("5078407bd58b175c5c225fdc"), "ns" : "mydb.foo", "min" : { "animal" : "cat" }, "max" : { "animal" : "dog" }, "shard" : "shard0004" }
42.2. Cong Database
547
These documents store the range of values for the shard key that describe the chunk in the min and max elds. Additionally the shard eld identies the shard in the cluster that owns the chunk. config.collections
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The collections (page 548) collection stores a document for each sharded collection in the cluster. Given a collection named pets in the records database, a document in the collections (page 548) collection would resemble the following:
{ "_id" : "records.pets", "lastmod" : ISODate("1970-01-16T15:00:58.107Z"), "dropped" : false, "key" : { "a" : 1 }, "unique" : false, "lastmodEpoch" : ObjectId("5078407bd58b175c5c225fdc") }
config.databases
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The databases (page 548) collection stores a document for each database in the cluster, and tracks if the database has sharding enabled. databases (page 548) represents each database in a distinct document. When a databases have sharding enabled, the primary eld holds the name of the primary shard.
{ "_id" : "admin", "partitioned" : false, "primary" : "config" } { "_id" : "mydb", "partitioned" : true, "primary" : "shard0000" }
config.lockpings
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The lockpings (page 548) collection keeps track of the active components in the sharded cluster. Given a cluster with a mongos (page 1032) running on example.com:30000, the document in the lockpings (page 548) collection would resemble:
{ "_id" : "example.com:30000:1350047994:16807", "ping" : ISODate("2012-10-12T18:32:54.892Z") }
config.locks
548
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The locks (page 548) collection stores a distributed lock. This ensures that only one mongos (page 1032) instance can perform administrative tasks on the cluster at once. The mongos (page 1032) acting as balancer takes a lock by inserting a document resembling the following into the locks collection.
{ "_id" : "balancer", "process" : "example.net:40000:1350402818:16807", "state" : 2, "ts" : ObjectId("507daeedf40e1879df62e5f3"), "when" : ISODate("2012-10-16T19:01:01.593Z"), "who" : "example.net:40000:1350402818:16807:Balancer:282475249", "why" : "doing balance round" }
If a mongos (page 1032) holds the balancer lock, the state eld has a value of 2, which means that balancer is active. The when eld indicates when the balancer began the current operation. Changed in version 2.0: The value of the state eld was 1 before MongoDB 2.0. config.mongos
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The mongos (page 549) collection stores a document for each mongos (page 1032) instance afliated with the cluster. mongos (page 1032) instances send pings to all members of the cluster every 30 seconds so the cluster can verify that the mongos (page 1032) is active. The ping eld shows the time of the last ping, while the up eld reports the uptime of the mongos (page 1032) as of the last ping. The cluster maintains this collection for reporting purposes. The following document shows the status of the mongos (page 1032) running on example.com:30000.
{ "_id" : "example.com:30000", "ping" : ISODate("2012-10-12T17:08:13.538Z"), "up" : 13699, "wait
config.settings
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The settings (page 549) collection holds the following sharding conguration settings: Chunk size. To change chunk size, see Modify Chunk Size (page 523). Balancer status. To change status, see Disable the Balancer (page 529). The following is an example settings collection:
{ "_id" : "chunksize", "value" : 64 } { "_id" : "balancer", "stopped" : false }
42.2. Cong Database
549
config.shards
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The shards (page 549) collection represents each shard in the cluster in a separate document. If the shard is a replica set, the host eld displays the name of the replica set, then a slash, then the hostname, as in the following example:
{ "_id" : "shard0000", "host" : "shard1/localhost:30000" }
If the shard has tags (page 532) assigned, this document has a tags eld, that holds an array of the tags, as in the following example:
{ "_id" : "shard0001", "host" : "localhost:30001", "tags": [ "NYC" ] }
config.tags
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The tags (page 550) collection holds documents for each tagged shard key range in the cluster. The documents in the tags (page 550) collection resemble the following:
{ "_id" : { "ns" : "records.users", "min" : { "zipcode" : "10001" } }, "ns" : "records.users", "min" : { "zipcode" : "10001" }, "max" : { "zipcode" : "10281" }, "tag" : "NYC" }
config.version
Internal MongoDB Metadata The config (page 545) database is internal: applications and administrators should not modify or depend upon its content in the course of normal operation. The version (page 550) collection holds the current metadata version number. This collection contains only one document: To access the version (page 550) collection you must use the db.getCollection() (page 978) method. For example, to display the collections document:
mongos> db.getCollection("version").find() { "_id" : 1, "version" : 3 }
Note: Like all databases in MongoDB, the config database contains a system.indexes (page 1099) collection contains metadata for all indexes in the database for information on indexes, see Indexes (page 327).
550
Part X
Application Development
551
MongoDB provides language-specic client libraries called drivers that let you develop applications to interact with your databases. This page lists the documents, tutorials, and reference pages that describe application development. For API-level documentation, see MongoDB Drivers and Client Libraries (page 555). For an overview application-development specic concepts, see the aggregation (page 253) and indexes (page 327) documents. For an introduction to basic MongoDB use, see the administration tutorials (page 123). See also: Core MongoDB Operations (CRUD) (page 165) section and the FAQ: MongoDB for Application Developers (page 713) document. Developers should also be familiar with the mongo shell and the MongoDB query and update operators (page 763).
553
554
CHAPTER 43
Development Considerations
The following documents outline basic application development documents:
43.1 MongoDB Drivers and Client Libraries

Applications communicate with MongoDB by way of a client library or driver that handles all interaction with the database in a language appropriate and sensible manner. See the following pages for more information about the MongoDB drivers: JavaScript (Language Center, docs) Python (Language Center, docs) Ruby (Language Center, docs) PHP (Language Center, docs) Perl (Language Center, docs) Java (Language Center, docs) Scala (Language Center, docs) C# (Language Center, docs) C (Language Center, docs) C++ (Language Center, docs) Haskell (Language Center, docs) Erlang (Language Center, docs)
43.2 Optimization Strategies for MongoDB

There are many factors that can affect database performance and responsiveness, including index use, query structure, data models and application design, as well as operational factors such as architecture and system conguration. This section describes techniques for optimizing application performance with MongoDB.
555
43.2.1 Optimize Query Performance with Indexes and Projections

Create Indexes to Support Queries For commonly issued queries, create indexes (page 327). If a query searches multiple elds, create a compound index (page 331). Scanning an index is much faster than scanning a collection. The indexes structures are smaller than the documents reference, and store references in order. Example If you have a posts collection containing blog posts, and if you regularly issue a query that sorts on the author_name eld, then you can optimize the query by creating an index on the author_name eld:
db.posts.ensureIndex( { author_name : 1 } )
Indexes also improve efciency on queries that routinely sort on a given eld. Example If you regularly issue a query that sorts on the timestamp eld, then you can optimize the query by creating an index on the timestamp eld: Creating this index:
db.posts.ensureIndex( { timestamp : 1 } )
Optimizes this query:

db.posts.find().sort( { timestamp : -1 } )
Because MongoDB can read indexes in both ascending and descending order, the direction of a single-key index does not matter. Indexes support queries, update operations, and some phases of the aggregation pipeline (page 257). Index keys that are of the BinData type are more efciently stored in the index if: the binary subtype value is in the range of 0-7 or 128-135, and the length of the byte array is: 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, or 32. Limit the Number of Query Results to Reduce Network Demand MongoDB cursors return results in groups of multiple documents. If you know the number of results you want, you can reduce the demand on network resources by issuing the cursor.limit() (page 959) method. This is typically used in conjunction with sort operations. For example, if you need only 10 results from your query to the posts collection, you would issue the following command:
db.posts.find().sort( { timestamp : -1 } ).limit(10)
For more information on limiting results, see cursor.limit() (page 959) Use Projections to Return Only Necessary Data When you need only a subset of elds from documents, you can achieve better performance by returning only the elds you need:
556
Chapter 43. Development Considerations
For example, if in your query to the posts collection, you need only the timestamp, title, author, and abstract elds, you would issue the following command:
db.posts.find( {}, { timestamp : 1 , title : 1 , author : 1 , abstract : 1} ).sort( { timestamp : -1
For more information on using projections, see Result Projections (page 171). Use $hint to Select a Particular Index In most cases the query optimizer (page 174) selects the optimal index for a specic operation; however, you can force MongoDB to use a specic index using the hint() (page 958) method. Use hint() (page 958) to support performance testing, or on some queries where you must select a eld or eld included in several indexes. Use the Increment Operator to Perform Operations Server-Side Use MongoDBs $inc (page 784) operator to increment or decrement values in documents. The operator increments the value of the eld on the server side, as an alternative to selecting a document, making simple modications in the client and then writing the entire document to the server. The $inc (page 784) operator can also help avoid race conditions, which would result when two application instances queried for a document, manually incremented a eld, and saved the entire document back at the same time.
43.2.2 Evaluate Performance of Current Operations

The following sections describe techniques for evaluating operational performance. Use the Database Proler to Evaluate Operations Against the Database MongoDB provides a database proler that shows performance characteristics of each operation against the database. Use the proler to locate any queries or write operations that are running slow. You can use this information, for example, to determine what indexes to create. For more information, see Database Proling (page 96). Use db.currentOp() to Evaluate mongod Operations The db.currentOp() (page 971) method reports on current operations running on a mongod (page 1021) instance. Use $explain to Evaluate Query Performance The explain() (page 953) method returns statistics on a query, and reports the index MongoDB selected to fulll the query, as well as information about the internal operation of the query. Example To use explain() (page 953) on a query for documents matching the expression { a: 1 }, in the collection named records, use an operation that resembles the following in the mongo (page 1036) shell:
db.records.find( { a: 1 } ).explain()
43.2. Optimization Strategies for MongoDB
557
43.2.3 Use Capped Collections for Fast Writes and Reads

Use Capped Collections for Fast Writes Capped Collections (page 558) are circular, xed-size collections that keep documents well-ordered, even without the use of an index. This means that capped collections can receive very high-speed writes and sequential reads. These collections are particularly useful for keeping log les but are not limited to that purpose. Use capped collections where appropriate. Use Natural Order for Fast Reads To return documents in the order they exist on disk, return sorted operations using the $natural (page 807) operator. On a capped collection, this also returns the documents in the order in which they were written. Natural order does not use indexes but can be fast for operations when you want to select the rst or last items on disk. See also: sort() (page 965) and limit() (page 959). See also: Server-side JavaScript (page 560).
43.3 Capped Collections

Capped collections are xed-size collections that support high-throughput operations that insert, retrieve, and delete documents based on insertion order. Capped collections work in a way similar to circular buffers: once a collection lls its allocated space, it makes room for new documents by overwriting the oldest documents in the collection. Capped collections have the following behaviors: Capped collections guarantee preservation of the insertion order. As a result, queries do not need an index to return documents in insertion order. Without this indexing overhead, they can support higher insertion throughput. Capped collections guarantee that insertion order is identical to the order on disk (natural order) and do so by prohibiting updates that increase document size. Capped collections only allow updates that t the original document size, which ensures a document does not change its location on disk. Capped collections automatically remove the oldest documents in the collection without requiring scripts or explicit remove operations. For example, the oplog.rs collection that stores a log of the operations in a replica set uses a capped collection. Consider the following potential uses cases for capped collections: Store log information generated by high-volume systems. Inserting documents in a capped collection without an index is close to the speed of writing log information directly to a le system. Furthermore, the built-in rst-in-rst-out property maintains the order of events, while managing storage use. Cache small amounts of data in a capped collections. Since caches are read rather than write heavy, you would either need to ensure that this collection always remains in the working set (i.e. in RAM) or accept some write penalty for the required index or indexes.
558
43.3.1 Recommendations and Restrictions

You cannot shard a capped collection. Capped collections created after 2.2 have an _id eld and an index on the _id eld by default. Capped collections created before 2.2 do not have an index on the _id eld by default. If you are using capped collections with replication prior to 2.2, you should explicitly create an index on the _id eld. You can update documents in a collection after inserting them; however, these updates cannot cause the documents to grow. If the update operation causes the document to grow beyond their original size, the update operation will fail. If you plan to update documents in a capped collection, remember to create an index to prevent update operations that require a table scan. You cannot delete documents from a capped collection. To remove all records from a capped collection, use the emptycapped command. To remove the collection entirely, use the drop() (page 921) method. Warning: If you have a capped collection in a replica set outside of the local database, before 2.2, you should create a unique index on _id. Ensure uniqueness using the unique: true option to the ensureIndex() (page 921) method or by using an ObjectId for the _id eld. Alternately, you can use the autoIndexId option to create (page 863) when creating the capped collection, as in the Query a Capped Collection (page 559) procedure. Use natural ordering to retrieve the most recently inserted elements from the collection efciently. This is (somewhat) analogous to tail on a log le.
43.3.2 Procedures
Create a Capped Collection You must create capped collections explicitly using the createCollection() (page 970) method, which is a helper in the mongo (page 1036) shell for the create (page 863) command. When creating a capped collection you must specify the maximum size of the collection in bytes, which MongoDB will pre-allocate for the collection. The size of the capped collection includes a small amount of space for internal overhead.
db.createCollection("mycoll", {capped:true, size:100000})
See also: db.createCollection() (page 970) and create (page 863). Query a Capped Collection If you perform a find() (page 924) on a capped collection with no ordering specied, MongoDB guarantees that the ordering of results is the same as the insertion order. To retrieve documents in reverse insertion order, issue find() (page 924) along with the sort() (page 965) method with the $natural (page 807) parameter set to -1, as shown in the following example:
db.cappedCollection.find().sort( { $natural: -1 } )
Check if a Collection is Capped Use the db.collection.isCapped() (page 937) method to determine if a collection is capped, as follows:
43.3. Capped Collections
559
db.collection.isCapped()
Convert a Collection to Capped You can convert a non-capped collection to a capped collection with the convertToCapped (page 861) command:
db.runCommand({"convertToCapped": "mycoll", size: 100000});
The size parameter species the size of the capped collection in bytes. Changed in version 2.2: Before 2.2, capped collections did not have an index on _id unless you specied autoIndexId to the create (page 863), after 2.2 this became the default. Automatically Remove Data After a Specied Period of Time For additional exibility when expiring data, consider MongoDBs TTL indexes, as described in Expire Data from Collections by Setting TTL (page 577). These indexes allow you to expire and remove data from normal collections using a special type, based on the value of a date-typed eld and a TTL value for the index. TTL Collections (page 577) are not compatible with capped collections. Tailable Cursor You can use a tailable cursor with capped collections. Similar to the Unix tail -f command, the tailable cursor tails the end of a capped collection. As new documents are inserted into the capped collection, you can use the tailable cursor to continue retrieving documents. See Create Tailable Cursor (page 568) for information on creating a tailable cursor.
43.4 Server-side JavaScript

Changed in version 2.4: The V8 JavaScript engine, which became the default in 2.4, allows multiple JavaScript operations to execute at the same time. Prior to 2.4, MongoDB operations that required the JavaScript interpreter had to acquire a lock, and a single mongod (page 1021) could only run a single JavaScript operation at a time.
43.4.1 Overview
MongoDB supports the execution of JavaScript code for the following server-side operations: mapReduce (page 814) and the corresponding mongo (page 1036) shell db.collection.mapReduce() (page 937). See Map-Reduce (page 311) for more information. method
eval (page 823) command, and the corresponding mongo (page 1036) shell method db.eval() (page 975) $where (page 775) operator Running .js les via a mongo shell Instance on the Server (page 561) JavaScript in MongoDB Although the above operations use JavaScript, most interactions with MongoDB do not use JavaScript but use an idiomatic driver (page 555) in the language of the interacting application.
560
See also: Store a JavaScript Function on the Server (page 561) You can disable all server-side execution of JavaScript, by passing the --noscripting (page 1026) option on the command line or setting noscripting (page 1084) in a conguration le.
43.4.2 Running .js les via a mongo shell Instance on the Server
You can run a JavaScript (.js) le using a mongo (page 1036) shell instance on the server. This is a good technique for performing batch administrative work. When you run mongo (page 1036) shell on the server, connecting via the localhost interface, the connection is fast with low latency.
43.4.3 Concurrency
Refer to the individual method or operator documentation for any concurrency information. See also the concurrency table (page 728).
43.5 Store a JavaScript Function on the Server

Note: We do not recommend using server-side stored functions if possible. There is a special system collection named system.js that can store JavaScript functions for reuse. To store a function, you can use the db.collection.save() (page 946), as in the following example:
db.system.js.save( { _id : "myAddFunction" , value : function (x, y){ return x + y; } } );
The _id eld holds the name of the function and is unique per database. The value eld holds the function denition Once you save a function in the system.js collection, you can use the function from any JavaScript context (e.g. eval (page 823) command or the mongo (page 1036) shell method db.eval() (page 975), $where (page 775) operator, mapReduce (page 814) or mongo (page 1036) shell method db.collection.mapReduce() (page 937)). Consider the following example from the mongo (page 1036) shell that rst saves a function named echoFunction to the system.js collection and calls the function using db.eval() (page 975) method:
db.system.js.save( { _id: "echoFunction", value : function(x) { return x; } } ) db.eval( "echoFunction( test )" )
See http://github.com/mongodb/mongo/tree/master/jstests/storefunc.js for a full example.
43.5. Store a JavaScript Function on the Server
561
New in version 2.1: In the mongo (page 1036) shell, you can use db.loadServerScripts() (page 982) to load all the scripts saved in the system.js collection for the current database. Once loaded, you can invoke the functions directly in the shell, as in the following example:
db.loadServerScripts(); echoFunction(3); myAddFunction(3, 5);
See also: Read Preference (page 402) Write Concern (page 398) Indexing Strategies (page 341) Aggregation Framework (page 255) Map-Reduce (page 311) Perform Incremental Map-Reduce (page 313) Troubleshoot the Map Function (page 317) Troubleshoot the Reduce Function (page 318) Connection String URI Format (page 1108)
562
CHAPTER 44
Application Design Patterns for MongoDB
The following documents provide patterns for developing application features:
44.1 Perform Two Phase Commits

44.1.1 Synopsis
This document provides a pattern for doing multi-document updates or transactions using a two-phase commit approach for writing data to multiple documents. Additionally, you can extend this process to provide a rollback (page 566) like functionality.
44.1.2 Background
Operations on a single document are always atomic with MongoDB databases; however, operations that involve multiple documents, which are often referred to as transactions, are not atomic. Since documents can be fairly complex and contain multiple nested documents, single-document atomicity provides necessary support for many practical use cases. Thus, without precautions, success or failure of the database operation cannot be all or nothing, and without support for multi-document transactions its possible for an operation to succeed for some operations and fail with others. When executing a transaction composed of several sequential operations the following issues arise: Atomicity: if one operation fails, the previous operation within the transaction must rollback to the previous state (i.e. the nothing, in all or nothing.) Isolation: operations that run concurrently with the transaction operation set must see a consistent view of the data throughout the transaction process. Consistency: if a major failure (i.e. network, hardware) interrupts the transaction, the database must be able to recover a consistent state. Despite the power of single-document atomic operations, there are cases that require multi-document transactions. For these situations, you can use a two-phase commit, to provide support for these kinds of multi-document updates. Because documents can represent both pending data and states, you can use a two-phase commit to ensure that data is consistent, and that in the case of an error, the state that preceded the transaction is recoverable (page 566). Note: Because only single-document operations are atomic with MongoDB, two-phase commits can only offer transaction-like semantics. Its possible for applications to return intermediate data at intermediate points during the two-phase commit or rollback.
563
44.1.3 Pattern
Overview The most common example of transaction is to transfer funds from account A to B in a reliable way, and this pattern uses this operation as an example. In a relational database system, this operation would encapsulate subtracting funds from the source (A) account and adding them to the destination (B) within a single atomic transaction. For MongoDB, you can use a two-phase commit in these situations to achieve a compatible response. All of the examples in this document use the mongo (page 1036) shell to interact with the database, and assume that you have two collections: First, a collection named accounts that will store data about accounts with one account per document, and a collection named transactions which will store the transactions themselves. Begin by creating two accounts named A and B, with the following command:
db.accounts.save({name: "A", balance: 1000, pendingTransactions: []}) db.accounts.save({name: "B", balance: 1000, pendingTransactions: []})
To verify that these operations succeeded, use find() (page 924):

db.accounts.find()
mongo (page 1036) will return two documents that resemble the following:
{ "_id" : ObjectId("4d7bc66cb8a04f512696151f"), "name" : "A", "balance" : 1000, "pendingTransactions" { "_id" : ObjectId("4d7bc67bb8a04f5126961520"), "name" : "B", "balance" : 1000, "pendingTransactions"
Transaction Description
Set Transaction State to Initial
Create the transaction collection by inserting the following document. The transaction document holds the source and destination, which refer to the name elds of the accounts collection, as well as the value eld that represents the amount of data change to the balance eld. Finally, the state eld reects the current state of the transaction.
db.transactions.save({source: "A", destination: "B", value: 100, state: "initial"})
To verify that these operations succeeded, use find() (page 924):

db.transactions.find()
This will return a document similar to the following:
{ "_id" : ObjectId("4d7bc7a8b8a04f5126961522"), "source" : "A", "destination" : "B", "value" : 100, "
Switch Transaction State to Pending
Before modifying either records in the accounts collection, set the transaction state to pending from initial. Set the local variable t in your shell session, to the transaction document using findOne() (page 929):
t = db.transactions.findOne({state: "initial"})
After assigning this variable t, the shell will return the value of t, you will see the following output:
564
Chapter 44. Application Design Patterns for MongoDB
{ "_id" : ObjectId("4d7bc7a8b8a04f5126961522"), "source" : "A", "destination" : "B", "value" : 100, "state" : "initial" }
Use update() (page 948) to change the value of state to pending:

db.transactions.update({_id: t._id}, {$set: {state: "pending"}}) db.transactions.find()
The find() (page 924) operation will return the contents of the transactions collection, which should resemble the following:
{ "_id" : ObjectId("4d7bc7a8b8a04f5126961522"), "source" : "A", "destination" : "B", "value" : 100, "
Apply Transaction to Both Accounts
Continue by applying the transaction to both accounts. The update() (page 948) query will prevent you from applying the transaction if the transaction is not already pending. Use the following update() (page 948) operation:
db.accounts.update({name: t.source, pendingTransactions: {$ne: t._id}}, {$inc: {balance: -t.value}, $ db.accounts.update({name: t.destination, pendingTransactions: {$ne: t._id}}, {$inc: {balance: t.value db.accounts.find()
The find() (page 924) operation will return the contents of the accounts collection, which should now resemble the following:
{ "_id" : ObjectId("4d7bc97fb8a04f5126961523"), "balance" : 900, "name" : "A", "pendingTransactions" { "_id" : ObjectId("4d7bc984b8a04f5126961524"), "balance" : 1100, "name" : "B", "pendingTransactions"
Set Transaction State to Committed
Use the following update() (page 948) operation to set the transactions state to committed:
db.transactions.update({_id: t._id}, {$set: {state: "committed"}}) db.transactions.find()
The find() (page 924) operation will return the contents of the transactions collection, which should now resemble the following:
{ "_id" : ObjectId("4d7bc7a8b8a04f5126961522"), "destination" : "B", "source" : "A", "state" : "commi
Remove Pending Transaction
Use the following update() (page 948) operation to set remove the pending transaction from the documents in the accounts collection:
db.accounts.update({name: t.source}, {$pull: {pendingTransactions: t._id}}) db.accounts.update({name: t.destination}, {$pull: {pendingTransactions: t._id}}) db.accounts.find()
44.1. Perform Two Phase Commits
565
The find() (page 924) operation will return the contents of the accounts collection, which should now resemble the following:
Set Transaction State to Done
Complete the transaction by setting the state of the transaction document to done:
db.transactions.update({_id: t._id}, {$set: {state: "done"}}) db.transactions.find()
The find() (page 924) operation will return the contents of the transactions collection, which should now resemble the following:
{ "_id" : ObjectId("4d7bc7a8b8a04f5126961522"), "destination" : "B", "source" : "A", "state" : "done"
Recovering from Failure Scenarios The most important part of the transaction procedure is not, the prototypical example above, but rather the possibility for recovering from the various failure scenarios when transactions do not complete as intended. This section will provide an overview of possible failures and provide methods to recover from these kinds of events. There are two classes of failures: all failures that occur after the rst step (i.e. setting the transaction set to initial (page 564)) but before the third step (i.e. applying the transaction to both accounts (page 565).) To recover, applications should get a list of transactions in the pending state and resume from the second step (i.e. switching the transaction state to pending (page 564).) all failures that occur after the third step (i.e. applying the transaction to both accounts (page 565)) but before the fth step (i.e. setting the transaction state to done (page 565).) To recover, application should get a list of transactions in the committed state and resume from the fourth step (i.e. remove the pending transaction (page 565).) Thus, the application will always be able to resume the transaction and eventually arrive at a consistent state. Run the following recovery operations every time the application starts to catch any unnished transactions. You may also wish run the recovery operation at regular intervals to ensure that your data remains consistent. The time required to reach a consistent state depends, on how long the application needs to recover each transaction.
Rollback
In some cases you may need to rollback or undo a transaction when the application needs to cancel the transaction, or because it can never recover as in cases where one of the accounts doesnt exist, or stops existing during the transaction. There are two possible rollback operations: 1. After you apply the transaction (page 565) (i.e. the third step,) you have fully committed the transaction and you should not roll back the transaction. Instead, create a new transaction and switch the values in the source and destination elds. 2. After you create the transaction (page 564) (i.e. the rst step,) but before you apply the transaction (page 565) (i.e the third step,) use the following process: 566 Chapter 44. Application Design Patterns for MongoDB
Set Transaction State to Canceling update() (page 948) operation:
Begin by setting the transactions state to canceling using the following
db.transactions.update({_id: t._id}, {$set: {state: "canceling"}})
Undo the Transaction counts:
Use the following sequence of operations to undo the transaction operation from both ac-
db.accounts.update({name: t.source, pendingTransactions: t._id}, {$inc: {balance: t.value}, $pull: {p db.accounts.update({name: t.destination, pendingTransactions: t._id}, {$inc: {balance: -t.value}, $pu db.accounts.find()
The find() (page 924) operation will return the contents of the accounts collection, which should resemble the following:
Set Transaction State to Canceled Finally, use the following update() (page 948) operation to set the transactions state to canceled: Step 3: set the transactions state to canceled:
db.transactions.update({_id: t._id}, {$set: {state: "canceled"}})
Multiple Applications
Transactions exist, in part, so that several applications can create and run operations concurrently without causing data inconsistency or conicts. As a result, it is crucial that only one 1 application can handle a given transaction at any point in time. Consider the following example, with a single transaction (i.e. T1) and two applications (i.e. A1 and A2). If both applications begin processing the transaction which is still in the initial state (i.e. step 1 (page 564)), then: A1 can apply the entire whole transaction before A2 starts. A2 will then apply T1 for the second time, because the transaction does not appear as pending in the accounts documents. To handle multiple applications, create a marker in the transaction document itself to identify the application that is handling the transaction. Use findAndModify() (page 926) method to modify the transaction:
t = db.transactions.findAndModify({query: {state: "initial", application: {$exists: 0}}, update: {$set: {state: "pending", application: "A1"}}, new: true})
When you modify and reassign the local shell variable t, the mongo (page 1036) shell will return the t object, which should resemble the following:
{ "_id" : ObjectId("4d7be8af2c10315c0847fc85"), "application" : "A1", "destination" : "B", "source" : "A", "state" : "pending", "value" : 150 }
44.1. Perform Two Phase Commits
567
Amend the transaction operations to ensure that only applications that match the identier in the value of the application eld before applying the transaction. If the application A1 fails during transaction execution, you can use the recovery procedures (page 566), but applications should ensure that they owns the transaction before applying the transaction. For example to resume pending jobs, use a query that resembles the following:
db.transactions.find({application: "A1", state: "pending"})
This will (or may) return a document from the transactions document that resembles the following:
{ "_id" : ObjectId("4d7be8af2c10315c0847fc85"), "application" : "A1", "destination" : "B", "source" :
44.1.4 Using Two-Phase Commits in Production Applications

The example transaction above is intentionally simple. For example, it assumes that: it is always possible roll back operations an account. account balances can hold negative values. Production implementations would likely be more complex. Typically accounts need to information about current balance, pending credits, pending debits. Then: when your application switches the transaction state to pending (page 564) (i.e. step 2) it would also make sure that the accounts has sufcient funds for the transaction. During this update operation, the application would also modify the values of the credits and debits as well as adding the transaction as pending. when your application removes the pending transaction (page 565) (i.e. step 4) the application would apply the transaction on balance, modify the credits and debits as well as removing the transaction from the pending eld., all in one update. Because all of the changes in the above two operations occur within a single update() (page 948) operation, these changes are all atomic. Additionally, for most important transactions, ensure that: the database interface (i.e. client library or driver) has a reasonable write concern congured to ensure that operations return a response on the success or failure of a write operation. your mongod (page 1021) instance has journaling enabled to ensure that your data is always in a recoverable state, in the event of an unclean mongod (page 1021) shutdown.
44.2 Create Tailable Cursor

44.2.1 Overview
By default, MongoDB will automatically close a cursor when the client has exhausted all results in the cursor. However, for capped collections (page 558) you may use a Tailable Cursor that remains open after the client exhausts the results in the initial cursor. Tailable cursors are conceptually equivalent to the tail Unix command with the -f option (i.e. with follow mode.) After clients insert new additional documents into a capped collection, the tailable cursor will continue to retrieve documents. Use tailable cursors on capped collections with high numbers of write operations for which an index would be too expensive. For instance, MongoDB replication (page 387) uses tailable cursors to tail the primarys oplog. Note: If your query is on an indexed eld, do not use tailable cursors, but instead, use a regular cursor. Keep track of
568
the last value of the indexed eld returned by the query. To retrieve the newly added documents, query the collection again using the last value of the indexed eld in the query criteria, as in the following example:
db.<collection>.find( { indexedField: { $gt: <lastvalue> } } )
Consider the following behaviors related to tailable cursors: Tailable cursors do not use indexes and return documents in natural order. Because tailable cursors do not use indexes, the initial scan for the query may be expensive; but, after initially exhausting the cursor, subsequent retrievals of the newly added documents are inexpensive. Tailable cursors may become dead, or invalid, if either: the query returns no match. the cursor returns the document at the end of the collection and then the application deletes those document. A dead cursor has an id of 0. See your driver documentation (page 555) for the driver-specic method to specify the tailable cursor. For more information on the details of specifying a tailable cursor, see MongoDB wire protocol documentation.
44.2.2 C++ Example

The tail function uses a tailable cursor to output the results from a query to a capped collection: The function handles the case of the dead cursor by having the query be inside a loop. To periodically check for new data, the cursor->more() statement is also inside a loop.
#include "client/dbclient.h" using namespace mongo; /* * Example of a tailable cursor. * The function "tails" the capped collection (ns) and output elements as they are added. * The function also handles the possibility of a dead cursor by tracking the field insertDate. * New documents are added with increasing values of insertDate. */ void tail(DBClientBase& conn, const char *ns) { BSONElement lastValue = minKey.firstElement(); Query query = Query().hint( BSON( "$natural" << 1 ) ); while ( 1 ) { auto_ptr<DBClientCursor> c = conn.query(ns, query, 0, 0, 0, QueryOption_CursorTailable | QueryOption_AwaitData ); while ( 1 ) { if ( !c->more() ) { if ( c->isDead() ) { break; }
44.2. Create Tailable Cursor
569
continue; } BSONObj o = c->next(); lastValue = o["insertDate"]; cout << o.toString() << endl; } query = QUERY( "insertDate" << GT << lastValue ).hint( BSON( "$natural" << 1 ) ); } }
The tail function performs the following actions: Initialize the lastValue variable, which tracks the last accessed value. The function will use the lastValue if the cursor becomes invalid and tail needs to restart the query. Use hint() (page 958) to ensure that the query uses the $natural (page 807) order. In an outer while(1) loop, Query the capped collection and return a tailable cursor that blocks for several seconds waiting for new documents
auto_ptr<DBClientCursor> c = conn.query(ns, query, 0, 0, 0, QueryOption_CursorTailable | QueryOption_AwaitData );
* Specify the capped collection using ns as an argument to the function. * Set the QueryOption_CursorTailable option to create a tailable cursor. * Set the QueryOption_AwaitData option so that the returned cursor blocks for a few seconds to wait for data. In an inner while (1) loop, read the documents from the cursor: * If the cursor has no more documents and is not invalid, loop the inner while loop to recheck for more documents. * If the cursor has no more documents and is dead, break the inner while loop. * If the cursor has documents: output the document, update the lastValue value, and loop the inner while (1) loop to recheck for more documents. If the logic breaks out of the inner while (1) loop and the cursor is invalid: * Use the lastValue value to create a new query condition that matches documents added after the lastValue. Explicitly ensure $natural order with the hint() method:
query = QUERY( "insertDate" << GT << lastValue ).hint( BSON( "$natural" << 1 ) );
* Loop through the outer while (1) loop to re-query with the new query condition and repeat. See also: Detailed blog post on tailable cursor
570
44.3 Isolate Sequence of Operations

44.3.1 Overview
Write operations are atomic on the level of a single document: no single write operation can atomically affect more than one document or more than one collection. When a single write operation modies multiple documents, the operation as a whole is not atomic, and other operations may interleave. The modication of a single document, or record, is always atomic, even if the write operation modies multiple sub-document within the single record. No other operations are atomic; however, you can isolate a single write operation that affects multiple documents using the isolation operator (page 796). This document describes one method of updating documents only if the local copy of the document reects the current state of the document in the database. In addition the following methods provide a way to manage isolated sequences of operations: the findAndModify() (page 926) provides an isolated query and modify operation. Perform Two Phase Commits (page 563) Create a unique index (page 334), to ensure that a key doesnt exist when you insert it.
44.3.2 Update if Current

In this pattern, you will: query for a document, modify the elds in that document and update the elds of a document only if the elds have not changed in the collection since the query. Consider the following example in JavaScript which attempts to update the qty eld of a document in the products collection:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
var myCollection = db.products; var myDocument = myCollection.findOne( { sku: abc123 } ); if (myDocument) { var oldQty = myDocument.qty; if (myDocument.qty < 10) { myDocument.qty *= 4; } else if ( myDocument.qty < 20 ) { myDocument.qty *= 3; } else { myDocument.qty *= 2; } myCollection.update( { _id: myDocument._id, qty: oldQty }, { $set: { qty: myDocument.qty }
44.3. Isolate Sequence of Operations
571
23 24 25 26 27 28 29 30 31 32 33 34
} ) var err = db.getLastErrorObj(); if ( err && err.code ) { print("unexpected error updating document: " + tojson( err )); } else if ( err.n == 0 ) { print("No update: no matching document for { _id: " + myDocument._id + ", qty: " + oldQty + " } }
Your application may require some modications of this pattern, such as: Use the entire document as the query in lines 18 and 19, to generalize the operation and guarantee that the original document was not modied, rather than ensuring that as single eld was not changed. Add a version variable to the document that applications increment upon each update operation to the documents. Use this version variable in the query expression. You must be able to ensure that all clients that connect to your database obey this constraint. Use $set (page 788) in the update expression to modify only your elds and prevent overriding other elds. Use one of the methods described in Create an Auto-Incrementing Sequence Field (page 572).
44.4 Create an Auto-Incrementing Sequence Field

44.4.1 Synopsis
MongoDB reserves the _id eld in the top level of all documents as a primary key. _id must be unique, and always has an index with a unique constraint (page 334). However, except for the unique constraint you can use any value for the _id eld in your collections. This tutorial describes two methods for creating an incrementing sequence number for the _id eld using the following: A Counters Collection (page 572) Optimistic Loop (page 574) Warning: Generally in MongoDB, you would not use an auto-increment pattern for the _id eld, or any eld, because it does not scale for databases with larger numbers of documents. Typically the default value ObjectId is more ideal for the _id.
A Counters Collection Use a separate counters collection tracks the last number sequence used. The _id eld contains the sequence name and the seq contains the last value of the sequence. 1. Insert into the counters collection, the initial value for the userid:
db.counters.insert( { _id: "userid", seq: 0 } )
572
2. Create a getNextSequence function that accepts a name of the sequence. The function uses the findAndModify() (page 926) method to atomically increment the seq value and return this new value:
function getNextSequence(name) { var ret = db.counters.findAndModify( { query: { _id: name }, update: { $inc: { seq: 1 } }, new: true } ); return ret.seq; }
3. Use this getNextSequence() function during insert() (page 936).

db.users.insert( { _id: getNextSequence("userid"), name: "Sarah C." } ) db.users.insert( { _id: getNextSequence("userid"), name: "Bob D." } )
You can verify the results with find() (page 924):

db.users.find()
The _id elds contain incrementing sequence values:

{ _id : 1, name : "Sarah C." } { _id : 2, name : "Bob D." }
Note: When findAndModify() (page 926) includes the upsert: true option and the query eld(s) is not uniquely indexed, the method could insert a document multiple times in certain circumstances. For instance, if multiple clients each invoke the method with the same query condition and these methods complete the nd phase before any of methods perform the modify phase, these methods could insert the same document. In the counters collection example, the query eld is the _id eld, which always has a unique index. Consider that the findAndModify() (page 926) includes the upsert: true option, as in the following modied example:
function getNextSequence(name) { var ret = db.counters.findAndModify( { query: { _id: name }, update: { $inc: { seq: 1 } }, new: true,
44.4. Create an Auto-Incrementing Sequence Field
573
upsert: true } ); return ret.seq; }
If multiple clients were to invoke the getNextSequence() method with the same name parameter, then the methods would observe one of the following behaviors: Exactly one findAndModify() (page 926) would successfully insert a new document. Zero or more findAndModify() (page 926) methods would update the newly inserted document. Zero or more findAndModify() (page 926) methods would fail when they attempted to insert a duplicate. If the method fails due to a unique index constraint violation, retry the method. Absent a delete of the document, the retry should not fail.
Optimistic Loop In this pattern, an Optimistic Loop calculates the incremented _id value and attempts to insert a document with the calculated _id value. If the insert is successful, the loop ends. Otherwise, the loop will iterate through possible _id values until the insert is successful. 1. Create a function named insertDocument that performs the insert if not present loop. The function wraps the insert() (page 936) method and takes a doc and a targetCollection arguments.
function insertDocument(doc, targetCollection) { while (1) { var cursor = targetCollection.find( {}, { _id: 1 } ).sort( { _id: -1 } ).limit(1); var seq = cursor.hasNext() ? cursor.next()._id + 1 : 1; doc._id = seq; targetCollection.insert(doc); var err = db.getLastErrorObj(); if( err && err.code ) { if( err.code == 11000 /* dup key */ ) continue; else print( "unexpected error inserting data: " + tojson( err ) ); } break; } }
The while (1) loop performs the following actions: Queries the targetCollection for the document with the maximum _id value. Determines the next sequence value for _id by:
574
adding 1 to the returned _id value if the returned cursor points to a document. otherwise: it sets the next sequence value to 1 if the returned cursor points to no document. For the doc to insert, set its _id eld to the calculated sequence value seq. Insert the doc into the targetCollection. If the insert operation errors with duplicate key, repeat the loop. Otherwise, if the insert operation encounters some other error or if the operation succeeds, break out of the loop. 2. Use the insertDocument() function to perform an insert:
var myCollection = db.users2; insertDocument( { name: "Grace H." }, myCollection ); insertDocument( { name: "Ted R." }, myCollection )
You can verify the results with find() (page 924):

db.users2.find()
The _id elds contain incrementing sequence values:

{ _id: 1, name: "Grace H." } { _id : 2, "name" : "Ted R." }
The while loop may iterate many times in collections with larger insert volumes.
44.5 Limit Number of Elements in an Array after an Update

New in version 2.4.
44.5.1 Synopsis
Consider an application where users may submit many scores (e.g. for a test), but the application only needs to track the top three test scores. This pattern uses the $push (page 792) operator with the $each (page 793), $sort (page 794), and $slice (page 793) modiers to sort and maintain an array of xed size.
44.5. Limit Number of Elements in an Array after an Update
575
Important: The array elements must be documents in order to use the $sort (page 794) modier.
44.5.2 Pattern
Consider the following document in the collection students:
{ _id: 1, scores: [ { attempt: 1, score: 10 }, { attempt: 2 , score:8 } ] }
The following update uses the $push (page 792) operator with: the $each (page 793) modier to append to the array 2 new elements, the $sort (page 794) modier to order the elements by ascending (1) score, and the $slice (page 793) modier to keep the last 3 elements of the ordered array.
db.students.update( { _id: 1 }, { $push: { scores: { $each : [ { attempt: 3, score: 7 }, { attempt: 4, score: 4 } ], $sort: { score: 1 }, $slice: -3 } } } )
Note: When using the $sort (page 794) modier on the array element, access the eld in the subdocument element directly instead of using the dot notation on the array eld. After the operation, the document contains the only the top 3 scores in the scores array:
{ "_id" : 1, "scores" : [ { "attempt" : 3, "score" : 7 }, { "attempt" : 2, "score" : 8 }, { "attempt" : 1, "score" : 10 } ] }
See also: $push (page 792) operator, $each (page 793) modier, $sort (page 794) modier, and $slice (page 793) modier.
576
44.6 Expire Data from Collections by Setting TTL

New in version 2.2. This document provides an introductions to MongoDBs time to live or TTL collection feature. Implemented as a special index type, TTL collections make it possible to store data in MongoDB and have the mongod (page 1021) automatically remove data after a specied period of time. This is ideal for some types of information like machine generated event data, logs, and session information that only need to persist in a database for a limited period of time.
44.6.1 Background
Collections expire by way of a special index that keeps track of insertion time in conjunction with a background thread in mongod (page 1021) that regularly removes expired documents from the collection. You can use this feature to expire data from replica sets and sharded clusters. Use the expireAfterSeconds option to the ensureIndex (page 921) method in conjunction with a TTL value in seconds to create an expiring collection. TTL collections set the usePowerOf2Sizes (page 857) collection ag, which means MongoDB must allocate more disk space relative to data size. This approach helps mitigate the possibility of storage fragmentation caused by frequent delete operations and leads to more predictable storage use patterns. Note: When the TTL thread is active, you will see a delete (page 227) operation in the output of db.currentOp() (page 971) or in the data collected by the database proler (page 99).
44.6.2 Constraints
Consider the following limitations: the indexed eld must be a date BSON type. If the eld does not have a date type, the data will not expire. you cannot create this index on the _id eld, or a eld that already has an index. the TTL index may not be compound (may not have multiple elds). if the eld holds an array, and there are multiple date-typed data in the index, the document will expire when the lowest (i.e. earliest) matches the expiration threshold. you cannot use a TTL index on a capped collection, because MongoDB cannot remove documents from a capped collection. Note: TTL indexes expire data by removing documents in a background task that runs every 60 seconds. As a result, the TTL index provides no guarantees that expired documents will not exist in the collection. Consider that: Documents may remain in a collection after they expire and before the background process runs. The duration of the removal operations depend on the workload of your mongod (page 1021) instance.
44.6.3 Enabling a TTL for a Collection

To set a TTL on the collection log.events for one hour use the following command at the mongo (page 1036) shell:
db.log.events.ensureIndex( { "status": 1 }, { expireAfterSeconds: 3600 } )
44.6. Expire Data from Collections by Setting TTL
577
The status eld must hold date/time information. MongoDB will automatically delete documents from this collection once the value of status is one or more hours old.
44.6.4 Replication
The TTL background thread only runs on primary members of replica sets. Secondaries members will replicate deletion operations from the primaries.
578
Part XI
The mongo Shell
579
The mongo (page 1036) shell is an interactive JavaScript shell for MongoDB, and is part of all MongoDB distributions. This section provides an introduction to the shell, and outlines key functions, operations, and use of the mongo (page 1036) shell. Most examples in the MongoDB Manual (page 1) use the mongo (page 1036) shell; however, many drivers (page 555) provide similar interfaces to MongoDB.
581
582
CHAPTER 45
Getting Started with the mongo Shell
This document provides a basic introduction to using the mongo (page 1036) shell. See Install MongoDB (page 3) for instructions on installing MongoDB for your system.
45.1 Start the mongo Shell

To start the mongo (page 1036) shell and connect to your MongoDB (page 1021) instance running on localhost with default port: 1. Go to your <mongodb installation dir>:
cd <mongodb installation dir>
2. Type ./bin/mongo to start mongo (page 1036):

./bin/mongo
If you have added the <mongodb installation dir>/bin to the PATH environment variable, you can just type mongo instead of ./bin/mongo. 3. To display the database you are using, type db:
db
The command should return test, which is the default database. To switch databases, issue the use <db> command, as in the following example:
use <database>
To list the available databases, use the command show dbs. See also How can I access different databases temporarily? (page 723) to access a different database from the current database without switching your current database context (i.e. db..) To start the mongo (page 1036) shell with other options, see examples of starting up mongo (page 1040) and mongo reference (page 1036) which provides details on the available options. Note: When starting, mongo (page 1036) checks the users HOME (page 1039) directory for a JavaScript le named .mongorc.js (page 1039). If found, mongo (page 1036) interprets the content of .mongorc.js before displaying the prompt for the rst time. If you use the shell to evaluate a JavaScript le or expression, either by using the --eval (page 1037) option on the command line or by specifying a .js le to mongo (page 1038), mongo (page 1036) will read the .mongorc.js le after the JavaScript has nished processing.
583
45.2 Executing Queries

From the mongo (page 1036) shell, you can use the shell methods (page 917) to run queries, as in the following example:
db.<collection>.find()
The db refers to the current database. The <collection> is the name of the collection to query. See Collection Help (page 592) to list the available collections. If the mongo (page 1036) shell does not accept the name of the collection, for instance if the name contains a space or starts with a number, you can use an alternate syntax to refer to the collection, as in the following:
db["3test"].find() db.getCollection("3test").find()
The find() (page 924) method is the JavaScript method to retrieve documents from <collection>. The find() (page 924) method returns a cursor to the results; however, in the mongo (page 1036) shell, if the returned cursor is not assigned to a variable, then the cursor is automatically iterated up to 20 times to print up to the rst 20 documents that match the query. The mongo (page 1036) shell will prompt Type it to iterate another 20 times. You can set the DBQuery.shellBatchSize attribute to change the number of iteration from the default value 20, as in the following example which sets it to 10:
DBQuery.shellBatchSize = 10;
For more information and examples on cursor handling in the mongo (page 1036) shell, see Cursors (page 175). See also Cursor Help (page 592) for list of cursor help in the mongo (page 1036) shell. For more documentation of basic MongoDB operations in the mongo (page 1036) shell, see: Getting Started with MongoDB (page 21) mongo Shell Quick Reference (page 597) Create (page 201) Read (page 209) Update (page 219) Delete (page 227) Indexing Operations (page 347) Read Operations (page 167) Write Operations (page 179)
45.3 Print
The mongo (page 1036) shell automatically prints the results of the find() (page 924) method if the returned cursor is not assigned to a variable. To format the result, you can add the .pretty() to the operation, as in the following:
db.<collection>.find().pretty()
584
Chapter 45. Getting Started with the mongo Shell
In addition, you can use the following explicit print methods in the mongo (page 1036) shell: print() to print without formatting print(tojson(<obj>)) to print with JSON formatting and equivalent to printjson() printjson() to print with JSON formatting and equivalent to print(tojson(<obj>))
45.4 Use a Custom Prompt

You may modify the content of the prompt by creating the variable prompt in the shell. The prompt variable can hold strings as well as any arbitrary JavaScript. If prompt holds a function that returns a string, mongo (page 1036) can display dynamic information in each prompt. Consider the following examples: Example Create a prompt with the number of commands issued in the current session, dene the following variables:
cmdCount = 1; prompt = function() { return (cmdCount++) + "> "; }
The prompt would then resemble the following:

1> db.collection.find() 2> show collections 3>
Example To create a mongo (page 1036) shell prompt in the form of <database>@<hostnane>$ dene the following variables:
host = db.serverStatus().host; prompt = function() { return db+"@"+host+"$ "; }

<database>@<hostname>$ use records switched to db records records@<hostname>$
Example To create a mongo (page 1036) shell prompt that contains the system up time and the number of documents in the current database, dene the following prompt variable:
prompt = function() { return "Uptime:"+db.serverStatus().uptime+" Documents:"+db.stats().objects+" > "; }
45.4. Use a Custom Prompt
585
Uptime:5897 Documents:6 > db.people.save({name : "James"}); Uptime:5948 Documents:7 >
45.5 Use an External Editor in the mongo Shell

New in version 2.2. In the mongo (page 1036) shell you can use the edit operation to edit a function or variable in an external editor. The edit operation uses the value of your environments EDITOR variable. At your system prompt you can dene the EDITOR variable and start mongo (page 1036) with the following two operations:
export EDITOR=vim mongo
Then, consider the following example shell session:

MongoDB shell version: 2.2.0 > function f() {} > edit f > f function f() { print("this really works"); } > f() this really works > o = {} { } > edit o > o { "soDoes" : "this" } >
Note: As mongo (page 1036) shell interprets code edited in an external editor, it may modify code in functions, depending on the JavaScript compiler. For mongo (page 1036) may convert 1+1 to 2 or remove comments. The actual changes affect only the appearance of the code and will vary based on the version of JavaScript used but will not affect the semantics of the code.
45.6 Exit the Shell

To exit the shell, type quit() or use the <Ctrl-c> shortcut.
586
Chapter 45. Getting Started with the mongo Shell
CHAPTER 46
Data Types in the mongo Shell
MongoDB BSON provide support for additional data types than JSON . Drivers (page 555) provide native support for these data types in host languages and the mongo (page 1036) shell also provides several helper classes to support the use of these data types in the mongo (page 1036) JavaScript shell. See MongoDB Extended JSON (page 1113) for additional information.
46.1 Date
The mongo (page 1036) shell provides various options to return the date, either as a string or as an object: Date() method which returns the current date as a string. Date() constructor which returns an ISODate object when used with the new operator. ISODate() constructor which returns an ISODate object when used with or without the new operator. Consider the following examples: To return the date as a string, use the Date() method, as in the following example:
var myDateString = Date();
To print the value of the variable, type the variable name in the shell, as in the following:
myDateString
The result is the value of myDateString:

Wed Dec 19 2012 01:03:25 GMT-0500 (EST)
To verify the type, use the typeof operator, as in the following:

typeof myDateString
The operation returns string. To get the date as an ISODate object, instantiate a new instance using the Date() constructor with the new operator, as in the following example:
var myDateObject = new Date();
To print the value of the variable, type the variable name in the shell, as in the following:
myDateObject
587
The result is the value of myDateObject:

ISODate("2012-12-19T06:01:17.171Z")

typeof myDateObject
The operation returns object. To get the date as an ISODate object, instantiate a new instance using the ISODate() constructor without the new operator, as in the following example:
var myDateObject2 = ISODate();
You can use the new operator with the ISODate() constructor as well. To print the value of the variable, type the variable name in the shell, as in the following:
myDateObject2
The result is the value of myDateObject2:

ISODate("2012-12-19T06:15:33.035Z")

typeof myDateObject2
The operation returns object.
46.2 ObjectId
The mongo (page 1036) shell provides the ObjectId() wrapper class around ObjectId data types. To generate a new ObjectId, use the following operation in the mongo (page 1036) shell:
new ObjectId
See also: ObjectId (page 194) for full documentation of ObjectIds in MongoDB.
46.3 NumberLong
By default, the mongo (page 1036) shell treats all numbers as oating-point values. The mongo (page 1036) shell provides the NumberLong() class to handle 64-bit integers. The NumberLong() constructor accepts the long as a string:
NumberLong("2090845886852")
The following examples use the NumberLong() class to write to the collection:
db.collection.insert( { _id: 10, calc: NumberLong("2090845886852") } ) db.collection.update( { _id: 10 }, { $set: { calc: NumberLong("2555555000000") } } ) db.collection.update( { _id: 10 }, { $inc: { calc: NumberLong(5) } } )
588
Chapter 46. Data Types in the mongo Shell
Retrieve the document to verify:

db.collection.findOne( { _id: 10 } )
In the returned document, the calc eld contains a NumberLong object:

{ "_id" : 10, "calc" : NumberLong("2555555000005") }
If you use the $inc (page 784) to increment the value of a eld that contains a NumberLong object by a oat, the data type changes to a oating point value, as in the following example: 1. Use $inc (page 784) to increment the calc eld by 5, which the mongo (page 1036) shell treats as a oat:
db.collection.update( { _id: 10 }, { $inc: { calc: 5 } } )
2. Retrieve the updated document:

db.collection.findOne( { _id: 10 } )
In the updated document, the calc eld contains a oating point value:
{ "_id" : 10, "calc" : 2555555000010 }
46.4 NumberInt
By default, the mongo (page 1036) shell treats all numbers as oating-point values. The mongo (page 1036) shell provides the NumberInt() constructor to explicitly specify 32-bit integers.
46.4. NumberInt
589
590
Chapter 46. Data Types in the mongo Shell
CHAPTER 47
Access the mongo Shell Help Information
In addition to the documentation in the MongoDB Manual (page 1), the mongo (page 1036) shell provides some additional information in its online help system. This document provides an overview of accessing this help information. See also: mongo Manual Page (page 1036) The mongo Shell (page 581) (MongoDB Manual (page 1) section on the shell.) mongo Shell Quick Reference (page 597).
47.1 Command Line Help

To see the list of options and help for starting the mongo (page 1036) shell, use the --help (page 1038) option from the command line:
mongo --help
47.2 Shell Help

To see the list of help, in the mongo (page 1036) shell, type help:
help
47.3 Database Help

To see the list of databases on the server, use the show dbs command:
show dbs
New in version 2.4: show databases is now an alias for show dbs To see the list of help for methods you can use on the db object, call the db.help() (page 980) method:
db.help()
591
To see the implementation of a method in the shell, type the db.<method name> without the parenthesis (()), as in the following example which will return the implementation of the method db.addUser() (page 967):
db.addUser
47.4 Collection Help

To see the list of collections in the current database, use the show collections command:
show collections
To see the help for methods available on the collection objects (e.g. db.<collection>.help() method:
db.collection.help()
db.<collection>), use the
<collection> can be the name of a collection that exists, although you may specify a collection that doesnt exist. To see the collection method implementation, type the db.<collection>.<method> name without the parenthesis (()), as in the following example which will return the implementation of the save() (page 946) method:
db.collection.save
47.5 Cursor Help

When you perform read operations (page 167) with the find() (page 924) method in the mongo (page 1036) shell, you can use various cursor methods to modify the find() (page 924) behavior and various JavaScript methods to handle the cursor returned from the find() (page 924) method. To list the available modier and cursor handling methods, use the db.collection.find().help() command:
db.collection.find().help()
<collection> can be the name of a collection that exists, although you may specify a collection that doesnt exist. To see the implementation of the cursor method, type the db.<collection>.find().<method> name without the parenthesis (()), as in the following example which will return the implementation of the toArray() method:
db.collection.find().toArray
Some useful methods for handling cursors are: hasNext() (page 958) which checks whether the cursor has more documents to return. next() (page 962) which returns the next document and advances the cursor position forward by one. forEach(<function>) (page 958) which iterates the whole cursor and applies the <function> to each document returned by the cursor. The <function> expects a single argument which corresponds to the document from each iteration.
592
Chapter 47. Access the mongo Shell Help Information
For examples on iterating a cursor and retrieving the documents from the cursor, see cursor handling (page 175). See also Cursor (page 951) for all available cursor methods.
47.6 Type Help

To get a list of the wrapper classes available in the mongo (page 1036) shell, such as BinData(), type help misc in the mongo (page 1036) shell:
help misc
47.6. Type Help
593
594
Chapter 47. Access the mongo Shell Help Information
CHAPTER 48
Write Scripts for the mongo Shell
You can write scripts for the mongo (page 1036) shell in JavaScript that manipulate data in MongoDB or perform administrative operation. For more information about the mongo (page 1036) shell see The mongo Shell (page 581), and see the Running .js les via a mongo shell Instance on the Server (page 561) section for more information about using these mongo (page 1036) script. This tutorial provides an introduction to writing JavaScript that uses the mongo (page 1036) shell to access MongoDB.
48.1 Opening New Connections

From the mongo (page 1036) shell or from a JavaScript le, you can instantiate database connections using the Mongo() (page 1007) constructor:
new Mongo() new Mongo(<host>) new Mongo(<host:port>)
Consider the following example that instantiates a new connection to the MongoDB instance running on localhost on the default port and sets the global db variable to myDatabase using the getDB() (page 1007) method:
conn = new Mongo(); db = conn.getDB("myDatabase");
Additionally, you can use the connect() method to connect to the MongoDB instance. The following example connects to the MongoDB instance that is running on localhost with the non-default port 27020 and set the global db variable:
db = connect("localhost:27020/myDatabase");
If you create new connections inside a JavaScript le (page 596): You cannot use use <dbname> inside the le to set the db global variable. To set the db global variable, use the getDB() (page 1007) method or the connect() method. You can assign the database reference to a variable other than db. Additionally, inside the script, you would need to call db.getLastErrorObj() (page 978) or db.getLastError() (page 978) explicitly to wait for the result of write operations (page 179).
595
48.2 Scripting
From the command line, use mongo (page 1036) to evaluate JavaScript.
48.2.1 --eval option

Use the --eval (page 1037) option to mongo (page 1036) to pass the shell a JavaScript fragment, as in the following:
mongo test --eval "printjson(db.getCollectionNames())"
This returns the output of db.getCollectionNames() (page 978) using the mongo (page 1036) shell connected to the mongod (page 1021) or mongos (page 1032) instance running on port 27017 on the localhost interface.
48.2.2 Evaluate a JavaScript le

You can specify a .js le to the mongo (page 1036) shell, and mongo (page 1036) will evaluate the javascript directly. Consider the following example:
mongo localhost:27017/test myjsfile.js
This operation evaluates the myjsfile.js script in a mongo (page 1036) shell that connects to the test database on the mongod (page 1021) instance accessible via the localhost interface on port 27017. Alternately, you can specify the mongodb connection parameters inside of the javascript le using the Mongo() constructor. See Opening New Connections (page 595) for more information.
596
Chapter 48. Write Scripts for the mongo Shell
CHAPTER 49
mongo Shell Quick Reference
49.1 mongo Shell Command History

You can retrieve previous commands issued in the mongo (page 1036) shell with the up and down arrow keys. Command history is stored in ~/.dbshell le. See .dbshell (page 1038) for more information.
49.2 Command Line Options

The mongo (page 1036) executable can be started with numerous options. See mongo executable (page 1036) page for details on all available options. The following table displays some common options for mongo (page 1036): Option --help (page 1038) --nodb (page 1036) --shell (page 1036) Description Show command line options Start mongo (page 1036) shell without connecting to a database. To connect later, see Opening New Connections (page 595). Used in conjunction with a JavaScript le (i.e. <le.js> (page 1038)) to continue in the mongo (page 1036) shell after running the JavaScript le. See JavaScript le (page 596) for an example.
49.3 Command Helpers

The mongo (page 1036) shell provides various help. The following table displays some common help methods and commands:
597
Help Methods and Description Commands help Show help. db.help() Show help for database methods. db.<collection>.help() Show help on collection methods. The <collection> can be the name of an existing collection or a non-existing collection. show dbs Print a list of all databases on the server. use <db> Switch current database to <db>. The mongo (page 1036) shell variable db is set to the current database. show Print a list of all collections for current database collections show users Print a list of users for current database. show profile Print the ve most recent operations that took 1 millisecond or more. See documentation on the database proler (page 99) for more information. show New in version 2.4: Print a list of all available databases. databases
49.4 Basic Shell JavaScript Operations

The mongo (page 1036) shell provides numerous mongo Shell Methods (page 917) methods for database operations. In the mongo (page 1036) shell, db is the variable that references the current database. The variable is automatically set to the default database test or is set when you use the use <db> to switch current database. The following table displays some common JavaScript operations:
598
Chapter 49. mongo Shell Quick Reference
JavaScript Database Operations db.auth() (page 968) coll = db.<collection>
Description If running in secure mode, authenticate the user. Set a specic collection in the current database to a variable coll, as in the following example: coll = db.myCollection; You can perform operations on the myCollection using the variable, as in the following example: coll.find(); Find all documents in the collection and returns a cursor. See the Read (page 209) and Read Operations (page 167) for more information and examples. See Cursors (page 175) for additional information on cursor handling in the mongo (page 1036) shell. Insert a new document into the collection. Update an existing document in the collection. See Update (page 219) for more information. Insert either a new document or update an existing document in the collection. See Update (page 219) for more information. Delete documents from the collection. See Delete (page 227) for more information. Drops or removes completely the collection. Create a new index on the collection if the index does not exist; otherwise, the operation has no effect. Return a reference to another database using this same connection without explicitly switching the current database. This allows for cross database queries. See How can I access different databases temporarily? (page 723) for more information.
db.collection.find() (page 924)
db.collection.insert() (page 936) db.collection.update() (page 948) db.collection.save() (page 946)
db.collection.remove() (page 944) db.collection.drop() (page 921) db.collection.ensureIndex() (page 921) db.getSiblingDB() db.getSisterDB() (page 980) or
For more information on performing operations in the shell, see: Create (page 201) Read (page 209) Update (page 219) Delete (page 227) Indexing Operations (page 347) Read Operations (page 167) Write Operations (page 179) mongo Shell Methods (page 917)
49.5 Keyboard Shortcuts

Changed in version 2.2. The mongo (page 1036) shell provides most keyboard shortcuts similar to those found in the bash shell or in Emacs. For some functions mongo (page 1036) provides multiple key bindings, to accommodate several familiar paradigms. The following table enumerates the keystrokes supported by the mongo (page 1036) shell:
49.5. Keyboard Shortcuts
599
Keystroke Up-arrow Down-arrow Home End Tab Left-arrow Right-arrow Ctrl-left-arrow Ctrl-right-arrow Meta-left-arrow Meta-right-arrow Ctrl-A Ctrl-B Ctrl-C Ctrl-D Ctrl-E Ctrl-F Ctrl-G Ctrl-J Ctrl-K Ctrl-L Ctrl-M Ctrl-N Ctrl-P Ctrl-R Ctrl-S Ctrl-T Ctrl-U Ctrl-W Ctrl-Y Ctrl-Z Ctrl-H (i.e. Backspace) Ctrl-I (i.e. Tab) Meta-B Meta-C Meta-D Meta-F Meta-L Meta-U Meta-Y Meta-[Backspace] Meta-< Meta->
Function previous-history next-history beginning-of-line end-of-line autocomplete backward-character forward-character backward-word forward-word backward-word forward-word beginning-of-line backward-char exit-shell delete-char (or exit shell) end-of-line forward-char abort accept-line kill-line clear-screen accept-line next-history previous-history reverse-search-history forward-search-history transpose-chars unix-line-discard unix-word-rubout yank Suspend (job control works in linux) backward-delete-char complete backward-word capitalize-word kill-word forward-word downcase-word upcase-word yank-pop backward-kill-word beginning-of-history end-of-history
49.6 Queries
In the mongo (page 1036) shell, perform read operations using the db.collection.find() (page 924) and db.collection.findOne() (page 929) methods. The db.collection.find() (page 924) method returns a cursor object which the mongo (page 1036) shell
600
iterates to print documents on screen. By default, mongo (page 1036) prints the rst 20. The mongo (page 1036) shell will prompt the user to Type it to continue iterating the next 20 results. The following table provides some common read operations in the mongo (page 1036) shell:
49.6. Queries
601
Read Operations db.collection.find(<query>) (page 924)
db.collection.find( <query>, <projection> ) (page 924)
db.collection.find().sort( <sort order> ) (page 965)
Description Find the documents matching the <query> criteria in the collection. If the <query> criteria is not specied or is empty (i.e {} ), the read operation selects all documents in the collection. The following example selects the documents in the users collection with the name eld equal to "Joe": coll = db.users; coll.find( { name: "Joe" } ); For more information on specifying the <query> criteria, see Query Document (page 168). Find documents matching the <query> criteria and return just specic elds in the <projection>. The following example selects all documents from the collection but returns only the name eld and the _id eld. The _id is always returned unless explicitly specied to not return. coll = db.users; coll.find( { }, { name: true } ); For more information on specifying the <projection>, see Result Projections (page 171). Return results in the specied <sort order>. The following example selects all documents from the collection and returns the results sorted by the name eld in ascending order (1). Use -1 for descending order: coll = db.users; coll.find().sort( { name: 1 } ); Return the documents matching the <query> criteria in the specied <sort order>. Limit result to <n> rows. Highly recommended if you need only a certain number of rows for best performance. Skip <n> results. Returns total number of documents in the collection. Returns the total number of documents that match the query. The count() (page 952) ignores limit() (page 959) and skip() (page 964). For example, if 100 records match but the limit is 10, count() (page 952) will return 100. This will be faster than iterating yourself, but still take time. Find and return a single document. Returns null if not found. The following example selects a single document in the users collection with the name eld matches to "Joe": coll = db.users; coll.findOne( { name: "Joe" } ); Internally, the findOne() (page 929) method is the find() (page 924) method with a limit(1) (page 959). Chapter 49. mongo Shell Quick Reference
db.collection.find( <query> ).sort( <sort order> ) (page 965) db.collection.find( ... ).limit( <n> ) (page 959) db.collection.find( ... ).skip( <n> ) (page 964) db.collection.count() (page 919) db.collection.find( <query> ).count() (page 952)
db.collection.findOne( <query> ) (page 929)
602
See Read (page 209) and Read Operations (page 167) documentation for more information and examples. See Query, Update and Projection Operators (page 763) to specify other query operators.
49.7 Error Checking Methods

The mongo (page 1036) shell provides numerous administrative database methods (page 917), including error checking methods. These methods are: Error Checking Methods db.getLastError() (page 978) db.getLastErrorObj() (page 978) Description Returns error message from the last operation. Returns the error document from the last operation.
49.8 Administrative Command Helpers

The following table lists some common methods to support database administration: JavaScript Database Administration Methods db.cloneDatabase(<host>) (page 969) db.copyDatabase(<from>, <to>, <host>) (page 970) Description
Clone the current database from the <host> specied. The <host> database instance must be in noauth mode. Copy the <from> database from the <host> to the <to> database on the current server. The <host> database instance must be in noauth mode. db.fromColl.renameCollection(<toColl>) Rename collection from fromColl to <toColl>. (page 945) db.repairDatabase() Repair and compact the current database. This operation can be very slow (page 984) on large databases. db.addUser( <user>, Add user to current database. <pwd> ) (page 967) db.getCollectionNames() Get the list of all collections in the current database. (page 978) db.dropDatabase() Drops the current database. (page 975) See also administrative database methods (page 917) for a full list of methods.
49.9 Opening Additional Connections

You can create new connections within the mongo (page 1036) shell. The following table displays the methods to create the connections: JavaScript Connection Create Methods db = connect("<host><:port>/<dbname>") conn = new Mongo() db = conn.getDB("dbname") Description Open a new database connection. Open a connection to a new server using new Mongo(). Use getDB() method of the connection to select a database.
49.7. Error Checking Methods
603
See also Opening New Connections (page 595) for more information on the opening new connections from the mongo (page 1036) shell.
49.10 Miscellaneous
The following table displays some miscellaneous methods: Method Object.bsonsize(<document>) Description Prints the BSON size of an <document>
See the MongoDB JavaScript API Documentation for a full list of JavaScript methods .
49.11 Additional Resources

Consider the following reference material that addresses the mongo (page 1036) shell and its interface: mongo (page 1036) mongo Shell Methods (page 917) Query, Update and Projection Operators (page 763) Database Commands (page 807) Aggregation Framework Reference (page 271) Additionally, the MongoDB source code repository includes a jstests directory which contains numerous mongo (page 1036) shell scripts. The Getting Started with MongoDB (page 21) provides a general introduction to MongoDB using examples from the mongo (page 1036) shell. Additionally, the following documents from other sections address areas relevant to users of the mongo (page 1036) shell: FAQ: The mongo Shell (page 723) mongo (page 1036) mongo Shell Methods (page 917) Furthermore, consider the following reference material that addresses the mongo (page 1036) shell and its interface: Query, Update and Projection Operators (page 763) Database Commands (page 807) Aggregation Framework Reference (page 271) Meta Query Operator Quick Reference (page 801)
604
Part XII
Use Cases
605
The use case documents introduce the patterns, designs, and operations used in application development with MongoDB. Each document provides concrete examples and implementation details to support core MongoDB use cases. These documents highlight application design, and data modeling strategies (i.e. schema design) for MongoDB with special attention to pragmatic considerations including indexing, performance, sharding, and scaling. Each document is distinct and can stand alone; however, each section builds on a set of common examples and general use cases. The operational intelligence case studies describe applications that collect machine generated data from logging systems, application output, and other systems. The product data management case studies address aspects of applications required for building product catalogs, and managing inventory in e-commerce systems. The content management case studies introduce basic patterns and techniques for building content management systems using MongoDB. Finally, the introductory application development tutorials with Python and MongoDB (page 675), provides a complete and fully developed application that you can build using MongoDB and popular Python web development tool kits.
607
608
CHAPTER 50
Operational Intelligence
As an introduction to the use of MongoDB for operational intelligence and real time analytics use, the document Storing Log Data (page 609) describes several ways and approaches to modeling and storing machine generated data with MongoDB. Then, Pre-Aggregated Reports (page 619) describes methods and strategies for processing data to generate aggregated reports from raw event-data. Finally Hierarchical Aggregation (page 628) presents a method for using MongoDB to process and store hierarchical reports (i.e. per-minute, per-hour, and per-day) from raw event data.
50.1 Storing Log Data

50.1.1 Overview
This document outlines the basic patterns and principles for using MongoDB as a persistent storage engine for log data from servers and other machine data. Problem Servers generate a large number of events (i.e. logging,) that contain useful information about their operation including errors, warnings, and users behavior. By default, most servers, store these data in plain text log les on their local le systems. While plain-text logs are accessible and human-readable, they are difcult to use, reference, and analyze without holistic systems for aggregating and storing these data. Solution The solution described below assumes that each server generates events also consumes event data and that each server can access the MongoDB instance. Furthermore, this design assumes that the query rate for this logging data is substantially lower than common for logging applications with a high-bandwidth event stream. Note: This case assumes that youre using an standard uncapped collection for this event data, unless otherwise noted. See the section on capped collections (page 619)
609
Schema Design The schema for storing log data in MongoDB depends on the format of the event data that youre storing. For a simple example, consider standard request logs in the combined format from the Apache HTTP Server. A line from these logs may resemble the following:
127.0.0.1 - frank [10/Oct/2000:13:55:36 -0700] "GET /apache_pb.gif HTTP/1.0" 200 2326 "[http://www.ex
The simplest approach to storing the log data would be putting the exact text of the log record into a document:
{
_id: ObjectId(4f442120eb03305789000000), line: 127.0.0.1 - frank [10/Oct/2000:13:55:36 -0700] "GET /apache_pb.gif HTTP/1.0" 200 2326 "[http: }
While this solution is does capture all data in a format that MongoDB can use, the data is not particularly useful, or its not terribly efcient: if you need to nd events that the same page, you would need to use a regular expression query, which would require a full scan of the collection. The preferred approach is to extract the relevant information from the log data into individual elds in a MongoDB document. When you extract data from the log into elds, pay attention to the data types you use to render the log data into MongoDB. As you design this schema, be mindful that the data types you use to encode the data can have a signicant impact on the performance and capability of the logging system. Consider the date eld: In the above example, [10/Oct/2000:13:55:36 -0700] is 28 bytes long. If you store this with the UTC timestamp type, you can convey the same information in only 8 bytes. Additionally, using proper types for your data also increases query exibility: if you store date as a timestamp you can make date range queries, whereas its very difcult to compare two strings that represent dates. The same issue holds for numeric elds; storing numbers as strings requires more space and is difcult to query. Consider the following document that captures all data from the above log entry:
{ _id: ObjectId(4f442120eb03305789000000), host: "127.0.0.1", logname: null, user: frank, time: ISODate("2000-10-10T20:55:36Z"), path: "/apache_pb.gif", request: "GET /apache_pb.gif HTTP/1.0", status: 200, response_size: 2326, referrer: "[http://www.example.com/start.html](http://www.example.com/start.html)", user_agent: "Mozilla/4.08 [en] (Win98; I ;Nav)" }
When extracting data from logs and designing a schema, also consider what information you can omit from your log tracking system. In most cases theres no need to track all data from an event log, and you can omit other elds. To continue the above example, here the most crucial information may be the host, time, path, user agent, and referrer, as in the following example document:
{ _id: ObjectId(4f442120eb03305789000000), host: "127.0.0.1", time: ISODate("2000-10-10T20:55:36Z"), path: "/apache_pb.gif", referer: "[http://www.example.com/start.html](http://www.example.com/start.html)",
610
Chapter 50. Operational Intelligence
user_agent: "Mozilla/4.08 [en] (Win98; I ;Nav)" }
You may also consider omitting explicit time elds, because the ObjectId embeds creation time:
{ _id: ObjectId(4f442120eb03305789000000), host: "127.0.0.1", path: "/apache_pb.gif", referer: "[http://www.example.com/start.html](http://www.example.com/start.html)", user_agent: "Mozilla/4.08 [en] (Win98; I ;Nav)" }
System Architecture The primary performance concern for event logging systems are: 1. how many inserts per second can it support, which limits the event throughput, and 2. how will the system manage the growth of event data, particularly concerning a growth in insert activity. In most cases the best way to increase the capacity of the system is to use an architecture with some sort of partitioning or sharding that distributes writes among a cluster of systems.
50.1.2 Operations
Insertion speed is the primary performance concern for an event logging system. At the same time, the system must be able to support exible queries so that you can return data from the system efciently. This section describes procedures for both document insertion and basic analytics queries. The examples that follow use the Python programming language and the PyMongo driver for MongoDB, but you can implement this system using any language you choose. Inserting a Log Record
Write Concern
MongoDB has a congurable write concern. This capability allows you to balance the importance of guaranteeing that all writes are fully recorded in the database with the speed of the insert. For example, if you issue writes to MongoDB and do not require that the database issue any response, the write operations will return very fast (i.e. asynchronously,) but you cannot be certain that all writes succeeded. Conversely, if you require that MongoDB acknowledge every write operation, the database will not return as quickly but you can be certain that every item will be present in the database. The proper write concern is often an application specic decision, and depends on the reporting requirements and uses of your analytics application.
Insert Performance
The following example contains the setup for a Python console session using PyMongo, with an event from the Apache Log:
50.1. Storing Log Data
611
>>> import bson >>> import pymongo >>> from datetime import datetime >>> conn = pymongo.MongoClient() >>> db = conn.event_db >>> event = { ... _id: bson.ObjectId(), ... host: "127.0.0.1", ... time: datetime(2000,10,10,20,55,36), ... path: "/apache_pb.gif", ... referer: "[http://www.example.com/start.html](http://www.example.com/start.html)", ... user_agent: "Mozilla/4.08 [en] (Win98; I ;Nav)" ...}
The following command will insert the event object into the events collection.
>>> db.events.insert(event, w=0)
By setting w=0, you do not require that MongoDB acknowledges receipt of the insert. Although very fast, this is risky because the application cannot detect network and server failures. See Write Concern (page 398) for more information. If you want to ensure that MongoDB acknowledges inserts, you can pass w=1 argument as follows:
>>> db.events.insert(event, w=1)
MongoDB also supports a more stringent level of write concern, if you have a lower tolerance for data loss: You can ensure that MongoDB not only acknowledge receipt of the message but also commit the write operation to the on-disk journal before returning successfully to the application, use can use the following insert() operation:
>>> db.events.insert(event, j=True)
Note: j=True implies w=1. Finally, if you have extremely low tolerance for event data loss, you can require that MongoDB replicate the data to multiple secondary replica set members before returning:
>>> db.events.insert(event, w=majority)
This will force your application to acknowledge that the data has replicated to a majority of congured members of the replica set. You can combine options as well:
>>> db.events.insert(event, j=True, w=majority)
In this case, your application will wait for a successful journal commit on the primary and a replication acknowledgment from a majority of congured secondaries. This is the safest option presented in this section, but it is the slowest. There is always a trade-off between safety and speed. Note: If possible, consider using bulk inserts to insert event data. All write concern options apply to bulk inserts, but you can pass multiple events to the insert() method at once. Batch inserts allow MongoDB to distribute the performance penalty incurred by more stringent write concern across a group of inserts. See also: Write Concern for Replica Sets (page 398) and getLastError (page 831).
612
Finding All Events for a Particular Page The value in maintaining a collection of event data derives from being able to query that data to answer specic questions. You may have a number of simple queries that you may use to analyze these data. As an example, you may want to return all of the events associated with specic value of a eld. Extending the Apache access log example from above, a common case would be to query for all events with a specic value in the path eld: This section contains a pattern for returning data and optimizing this operation.
Query
Use a query that resembles the following to return all documents http://docs.mongodb.org/manual/apache_pb.gif value in the path eld:
>>> q_events = db.events.find({path: /apache_pb.gif})
with
the
Note: If you choose to shard the collection that stores this data, the shard key you choose can impact the performance of this query. See the sharding (page 617) section of the sharding document.
Index Support
Adding an index on the path eld would signicantly enhance the performance of this operation.
>>> db.events.ensure_index(path)
Because the values of the path likely have a random distribution, in order to operate efciently, the entire index should be resident in RAM. In this case, the number of distinct paths is typically small in relation to the number of documents, which will limit the space that the index requires. If your system has a limited amount of RAM, or your data set has a wider distribution in values, you may need to re investigate your indexing support. In most cases, however, this index is entirely sufcient. See also: The db.collection.ensureIndex() (page db.events.ensure_index() method in PyMongo. Finding All the Events for a Particular Date The next example describes the process for returning all the events for a particular date.
Query
921)
JavaScript
method
and
the
To retrieve this data, use the following query:

>>> q_events = db.events.find(time: ... { $gte:datetime(2000,10,10),$lt:datetime(2000,10,11)})
Index Support
In this case, an index on the time eld would optimize performance:
613
>>> db.events.ensure_index(time)
Because your application is inserting events in order, the parts of the index that capture recent events will always be in active RAM. As a result, if you query primarily on recent data, MongoDB will be able to maintain a large index, quickly fulll queries, and avoid using much system memory. See also: The db.events.ensureIndex() (page 921) JavaScript method and the db.events.ensure_index() method in PyMongo. Finding All Events for a Particular Host/Date The following example describes a more complex query for returning all events in the collection for a particular host on a particular date. This kinds analysis may be useful for investigating suspicious behavior by a specic user.
Query
Use a query that resembles the following:

>>> q_events = db.events.find({ ... host: 127.0.0.1, ... time: {$gte:datetime(2000,10,10),$lt:datetime(2000,10,11)} ... })
This query selects documents from the events collection where the host eld is 127.0.0.1 (i.e. local host), and the value of the time eld represents a date that is on or after (i.e. $gte (page 765)) 2000-10-10 but before (i.e. $lt (page 766)) 2000-10-11.
Index Support
The indexes you use may have signicant implications for the performance of these kinds of queries. For instance, you can create a compound index on the time and host eld, using the following command:
>>> db.events.ensure_index([(time, 1), (host, 1)])
To analyze the performance for the above query using this index, issue the q_events.explain() method in a Python console. This will return something that resembles:
{ ... ucursor: uBtreeCursor time_1_host_1, uindexBounds: {uhost: [[u127.0.0.1, u127.0.0.1]], utime: [ [ datetime.datetime(2000, 10, 10, 0, 0), datetime.datetime(2000, 10, 11, 0, 0)]] }, ... umillis: 4, un: 11, unscanned: 1296, unscannedObjects: 11, ... }
This query had to scan 1296 items from the index to return 11 objects in 4 milliseconds. Conversely, you can test a different compound index with the host eld rst, followed by the time eld. Create this index using the following operation: 614 Chapter 50. Operational Intelligence
>>> db.events.ensure_index([(host, 1), (time, 1)])
Use the q_events.explain() operation to test the performance:

{ ... ucursor: uBtreeCursor host_1_time_1, uindexBounds: {uhost: [[u127.0.0.1, u127.0.0.1]], utime: [[datetime.datetime(2000, 10, 10, 0, 0), datetime.datetime(2000, 10, 11, 0, 0)]]}, ... umillis: 0, un: 11, ... unscanned: 11, unscannedObjects: 11, ... }
Here, the query had to scan 11 items from the index before returning 11 objects in less than a millisecond. By placing the more selective element of your query rst in a compound index you may be able to build more useful queries. Note: Although the index order has an impact query performance, remember that index scans are much faster than collection scans, and depending on your other queries, it may make more sense to use the { time: 1, host: 1 } index depending on usage prole. See also: The db.events.ensureIndex() (page 921) JavaScript method and the db.events.ensure_index() method in PyMongo. Counting Requests by Day and Page The following example describes the process for using the collection of Apache access events to determine the number of request per resource (i.e. page) per day in the last month.
Aggregation
New in version 2.1. The aggregation framework provides the capacity for queries that select, process, and aggregate results from large numbers of documents. The aggregate() (page 918) offers greater exibility, capacity with less complexity than the existing mapReduce (page 814) and group (page 810) aggregation commands. Consider the following aggregation pipeline:
1
>>> result = db.command(aggregate, events, pipeline=[ ... { $match: { ... time: { ... $gte: datetime(2000,10,1), ... $lt: datetime(2000,11,1) } } }, ... { $project: { ... path: 1, ... date: { ... y: { $year: $time },
1 To translate statements from the aggregation framework (page 255) to SQL, you can consider the $match (page 281) equivalent to WHERE, $project (page 285) to SELECT, and $group (page 278) to GROUP BY.
615
... ... ... ... ... ... ... ... ... ...
m: { $month: $time }, d: { $dayOfMonth: $time } } } }, { $group: { _id: { p:$path, y: $date.y, m: $date.m, d: $date.d }, hits: { $sum: 1 } } }, ])
This command aggregates documents from the events collection with a pipeline that: 1. Uses the $match (page 281) to limit the documents that the aggregation framework must process. $match (page 281) is similar to a find() (page 924) query. This operation selects all documents where the value of the time eld represents a date that is on or after (i.e. $gte (page 765)) 2000-10-10 but before (i.e. $lt (page 766)) 2000-10-11. 2. Uses the $project (page 285) to limit the data that continues through the pipeline. This operator: Selects the path eld. Creates a y eld to hold the year, computed from the time eld in the original documents. Creates a m eld to hold the month, computed from the time eld in the original documents Creates a d eld to hold the day, computed from the time eld in the original documents. 3. Uses the $group (page 278) to create new computed documents. This step will create a single new document for each unique path/date combination. The documents take the following form: the _id eld holds a sub-document with the contents path eld from the original documents in the p eld, with the date elds from the $project (page 285) as the remaining elds. the hits eld use the $sum (page 289) statement to increment a counter for every document in the group. In the aggregation output, this eld holds the total number of documents at the beginning of the aggregation pipeline with this unique date and path. Note: In sharded environments, the performance of aggregation operations depends on the shard key. Ideally, all the items in a particular $group (page 278) operation will reside on the same server. While this distribution of documents would occur if you chose the time eld as the shard key, a eld like path also has this property and is a typical choice for sharding. Also see the sharding considerations (page 617). of this document for additional recommendations for using sharding. See also: Aggregation Framework (page 255)
Index Support
To optimize the aggregation operation, ensure that the initial $match (page 281) query has an index. Use the following command to create an index on the time eld in the events collection:
>>> db.events.ensure_index(time)
Note: If you have already created a compound index on the time and host (i.e. { time: 1, host, 1 },) MongoDB will use this index for range queries on just the time eld. Do not create an additional index, in these situations. 616 Chapter 50. Operational Intelligence
50.1.3 Sharding
Eventually your systems events will exceed the capacity of a single event logging database instance. In these situations you will want to use a sharded cluster, which takes advantage of MongoDBs sharding functionality. This section introduces the unique sharding concerns for this event logging case. See also: Sharding (page 483) and FAQ: Sharding with MongoDB (page 731) Limitations In a sharded environment the limitations on the maximum insertion rate are: the number of shards in the cluster. the shard key you chose. Because MongoDB distributed data in using ranges (i.e. chunks) of keys, the choice of shard key can control how MongoDB distributes data and the resulting systems capacity for writes and queries. Ideally, your shard key should allow insertions balance evenly among the shards 2 and for most queries to only need to access a single shard. 3 Continue reading for an analysis of a collection of shard key choices. Shard by Time While using the timestamp, or the ObjectId in the _id eld, these keys lead to two problems:
4
would distribute your data evenly among shards,
1. All inserts always ow to the same shard, which means that your sharded cluster will have the same write throughput as a standalone instance. 2. Most reads will tend to cluster on the same shard, as analytics queries. Shard by a Semi-Random Key To distribute data more evenly among the shards, you may consider using a more random piece of data, such as a hash of the _id eld (i.e. the ObjectId as a shard key. While this introduces some additional complexity into your application, to generate the key, it will distribute writes among the shards. In these deployments having 5 shards will provide 5 times the write capacity as a single instance. Using this shard key, or any hashed value as a key presents the following downsides: the shard key, and the index on the key will consume additional space in the database. queries, unless they include the shard key itself, 5 must run in parallel on all shards, which may lead to degraded performance. This might be an acceptable trade-off in some situations. The workload of event logging systems tends to be heavily skewed toward writing, read performance may not be as critical as more robust write performance.
2 For this reason, avoid shard keys based on the timestamp or the insertion time (i.e. the ObjectId) because all writes will end up on a single node. 3 For this reason, avoid randomized shard keys (e.g. hash based shard keys) because any query will have to access all shards in the cluster. 4 The ObjectId derives from the creation time, and is effectively a timestamp in this case. 5 Typically, it is difcult to use these kinds of shard keys in queries.
617
Shard by an Evenly-Distributed Key in the Data Set If a eld in your documents has values that are evenly distributed among the documents, you may consider using this key as a shard key. Continuing the example from above, you may consider using the path eld. Which may have a couple of advantages: 1. writes will tend to balance evenly among shards. 2. reads will tend to be selective and local to a single shard if the query selects on the path eld. There are a few potential problems with these kinds of shard keys: 1. If a large number of documents will have the same shard key, you run the risk of having a portion of your data collection MongoDB cannot distribute throughout the cluster. 2. If there are a small number of possible values, there may be a limit to how much MongoDB will be able to distribute the data among the shard. Note: Test using your existing data to ensure that the distribution is truly even, and that there is a sufcient quantity of distinct values for the shard key.
Shard by Combine a Natural and Synthetic Key MongoDB supports compound shard keys that combine the best aspects of sharding by a evenly distributed key in the set (page 618) and sharding by a random key (page 617). In these situations, the shard key would resemble { path: 1 , ssk: 1 } where, path is an often used natural key, or value from your data and ssk is a hash of the _id eld. 6 Using this type of shard key, data is largely distributed by the natural key, or path, which makes most queries that access the path eld local to a single shard or group of shards. At the same time, if there is not sufcient distribution for specic values of path, the ssk makes it possible for MongoDB to create chunks and data across the cluster. In most situations, these kinds of keys provide the ideal balance between distributing writes across the cluster and ensuring that most queries will only need to access a select number of shards. Test with Your Own Data Selecting shard keys is difcult because: there are no denitive best-practices, the decision has a large impact on performance, and it is difcult or impossible to change the shard key after making the selection. The sharding options (page 617) provides a good starting point for thinking about shard key selection. Nevertheless, the best way to select a shard key is to analyze the actual insertions and queries from your own application.
50.1.4 Managing Event Data Growth

Without some strategy for managing the size of your database, most event logging systems can grow innitely. This is particularly important in the context of MongoDB may not relinquish data to the le system in the way you might expect. Consider the following strategies for managing data growth:
6
You must still calculate the value of this synthetic key in your application when you insert documents into your collection.
618
Capped Collections Depending on your data retention requirements as well as your reporting and analytics needs, you may consider using a capped collection to store your events. Capped collections have a xed size, and drop old data when inserting new data after reaching cap. Note: In the current version, it is not possible to shard capped collections.
Multiple Collections, Single Database Strategy: Periodically rename your event collection so that your data collection rotates in much the same way that you might rotate log les. When needed, you can drop the oldest collection from the database. This approach has several advantages over the single collection approach: 1. Collection renames are fast and atomic. 2. MongoDB does not bring any document into memory to drop a collection. 3. MongoDB can effectively reuse space freed by removing entire collections without leading to data fragmentation. Nevertheless, this operation may increase some complexity for queries, if any of your analyses depend on events that may reside in the current and previous collection. For most real time data collection systems, this approach is the most ideal. Multiple Databases Strategy: Rotate databases rather than collections, as in the Multiple Collections, Single Database (page 619) example. While this signicantly increases application complexity for insertions and queries, when you drop old databases, MongoDB will return disk space to the le system. This approach makes the most sense in scenarios where your event insertion rates and/or your data retention rates were extremely variable. For example, if you are performing a large backll of event data and want to make sure that the entire set of event data for 90 days is available during the backll, during normal operations you only need 30 days of event data, you might consider using multiple databases.
50.2 Pre-Aggregated Reports

50.2.1 Overview
This document outlines the basic patterns and principles for using MongoDB as an engine for collecting and processing events in real time for use in generating up to the minute or second reports. Problem Servers and other systems can generate a large number of documents, and it can be difcult to access and analyze such large collections of data originating from multiple servers. This document makes the following assumptions about real-time analytics:
50.2. Pre-Aggregated Reports
619
There is no need to retain transactional event data in MongoDB, and how your application handles transactions is outside of the scope of this document. You require up-to-the minute data, or up-to-the-second if possible. The queries for ranges of data (by time) must be as fast as possible. See also: Storing Log Data (page 609). Solution The solution described below assumes a simple scenario using data from web server access logs. With this data, you will want to return the number of hits to a collection of web sites at various levels of granularity based on time (i.e. by minute, hour, day, week, and month) as well as by the path of a resource. To achieve the required performance to support these tasks, upserts and increment (page 784) operations will allow you to calculate statistics, produce simple range-based queries, and generate lters to support time-series charts of aggregated data.
50.2.2 Schema
Schemas for real-time analytics systems must support simple and fast query and update operations. In particular, attempt to avoid the following situations which can degrade performance: documents growing signicantly after creation. Document growth forces MongoDB to move the document on disk, which can be time and resource consuming relative to other operations; queries requiring MongoDB to scan documents in the collection without using indexes; and deeply nested documents that make accessing particular elds slow. Intuitively, you may consider keeping hit counts in individual documents with one document for every unit of time (i.e. minute, hour, day, etc.) However, queries must return multiple documents for all non-trivial time-rage queries, which can slow overall query performance. Preferably, to maximize query performance, use more complex documents, and keep several aggregate values in each document. The remainder of this section outlines several schema designs that you may consider for this real-time analytics system. While there is no single pattern for every problem, each pattern is more well suited to specic classes of problems. One Document Per Page Per Day Consider the following example schema for a solution that stores all statistics for a single day and page in a single document:
{ _id: "20101010/site-1/apache_pb.gif", metadata: { date: ISODate("2000-10-10T00:00:00Z"), site: "site-1", page: "/apache_pb.gif" }, daily: 5468426, hourly: { "0": 227850,
620
"1": 210231, ... "23": 20457 }, minute: { "0": 3612, "1": 3241, ... "1439": 2819 } }
This approach has a couple of advantages: For every request on the website, you only need to update one document. Reports for time periods within the day, for a single page require fetching a single document. There are, however, signicant issues with this approach. The most signicant issue is that, as you upsert data into the hourly and monthly elds, the document grows. Although MongoDB will pad the space allocated to documents, it must still will need to reallocate these documents multiple times throughout the day, which impacts performance. Pre-allocate Documents
Simple Pre-Allocation
To mitigate the impact of repeated document migrations throughout the day, you can tweak the one document per page per day (page 620) approach by adding a process that pre-allocates documents with elds that hold 0 values throughout the previous day. Thus, at midnight, new documents will exist. Note: To avoid situations where your application must pre-allocate large numbers of documents at midnight, its best to create documents throughout the previous day by upserting randomly when you update a value in the current days data. This requires some tuning, to balance two requirements: 1. your application should have pre-allocated all or nearly all of documents by the end of the day. 2. your application should infrequently pre-allocate a document that already exists to save time and resources on extraneous upserts. As a starting point, consider the average number of hits a day (h), and then upsert a blank document upon update with a probability of 1/h. Pre-allocating increases performance by initializing all documents with 0 values in all elds. After create, documents will never grow. This means that: 1. there will be no need to migrate documents within the data store, which is a problem in the one document per page per day (page 620) approach. 2. MongoDB will not add padding to the records, which leads to a more compact data representation and better memory use of your memory.
Add Intra-Document Hierarchy
Note: MongoDB stores BSON documents as a sequence of elds and values, not as a hash table. As a result, writing to the eld stats.mn.0 is considerably faster than writing to stats.mn.1439.
621
Figure 50.1: In order to update the value in minute #1349, MongoDB must skip over all 1349 entries before it. To optimize update and insert operations you can introduce intra-document hierarchy. In particular, you can split the minute eld up into 24 hourly elds:
{ _id: "20101010/site-1/apache_pb.gif", metadata: { date: ISODate("2000-10-10T00:00:00Z"), site: "site-1", page: "/apache_pb.gif" }, daily: 5468426, hourly: { "0": 227850, "1": 210231, ... "23": 20457 }, minute: { "0": { "0": 3612, "1": 3241, ... "59": 2130 }, "1": { "60": ... , }, ... "23": { ... "1439": 2819 } } }
This allows MongoDB to skip forward throughout the day when updating the minute data, which makes the update performance more uniform and faster later in the day.
Figure 50.2: To update the value in minute #1349, MongoDB rst skips the rst 23 hours and then skips 59 minutes for only 82 skips as opposed to 1439 skips in the previous schema.
Separate Documents by Granularity Level Pre-allocating documents (page 621) is a reasonable design for storing intra-day data, but the model breaks down when displaying data over longer multi-day periods like months or quarters. In these cases, consider storing daily statistics in a single document as above, and then aggregate monthly data into a separate document.
622
This introduce a second set of upsert operations to the data collection and aggregation portion of your application but the gains reduction in disk seeks on the queries, should be worth the costs. Consider the following example schema: 1. Daily Statistics
{ _id: "20101010/site-1/apache_pb.gif", metadata: { date: ISODate("2000-10-10T00:00:00Z"), site: "site-1", page: "/apache_pb.gif" }, hourly: { "0": 227850, "1": 210231, ... "23": 20457 }, minute: { "0": { "0": 3612, "1": 3241, ... "59": 2130 }, "1": { "0": ..., }, ... "23": { "59": 2819 } } }
2. Monthly Statistics
{ _id: "201010/site-1/apache_pb.gif", metadata: { date: ISODate("2000-10-00T00:00:00Z"), site: "site-1", page: "/apache_pb.gif" }, daily: { "1": 5445326, "2": 5214121, ... } }
50.2.3 Operations
This section outlines a number of common operations for building and interacting with real-time-analytics reporting system. The major challenge is in balancing performance and write (i.e. upsert) performance. All examples in this document use the Python programming language and the PyMongo driver for MongoDB, but you can implement this system using any language you choose. Log an Event Logging an event such as a page request (i.e. hit) is the main write activity for your system. To maximize performance, youll be doing in-place updates with the upsert operation. Consider the following example:
623
from datetime import datetime, time def log_hit(db, dt_utc, site, page): # Update daily stats doc id_daily = dt_utc.strftime(%Y%m%d /) + site + page hour = dt_utc.hour minute = dt_utc.minute # Get a datetime that only includes date info d = datetime.combine(dt_utc.date(), time.min) query = { _id: id_daily, metadata: { date: d, site: site, page: page } } update = { $inc: { hourly.%d % (hour,): 1, minute.%d .%d % (hour,minute): 1 } } db.stats.daily.update(query, update, upsert=True) # Update monthly stats document id_monthly = dt_utc.strftime(%Y%m/) + site + page day_of_month = dt_utc.day query = { _id: id_monthly, metadata: { date: d.replace(day=1), site: site, page: page } } update = { $inc: { daily.%d % day_of_month: 1} } db.stats.monthly.update(query, update, upsert=True)
The upsert operation (i.e. upsert=True) performs an update if the document exists, and an insert if the document does not exist. Note: This application requires upserts, because the pre-allocation (page 624) method only pre-allocates new documents with a high probability, not with complete certainty. Without preallocation, you end up with a dynamically growing document, slowing upserts as MongoDB moves documents to accommodate growth.
Pre-allocate To prevent document growth, you can preallocate new documents before the system needs them. As you create new documents, set all values to 0 for so that documents will not grow to accommodate updates. Consider the following preallocate() function:
def preallocate(db, dt_utc, site, page): # Get id values id_daily = dt_utc.strftime(%Y%m%d /) + site + page id_monthly = dt_utc.strftime(%Y%m/) + site + page # Get daily metadata daily_metadata = { date: datetime.combine(dt_utc.date(), time.min),
624
site: site, page: page } # Get monthly metadata monthly_metadata = { date: daily_metadata[date].replace(day=1), site: site, page: page } # Initial zeros for statistics hourly = dict((str(i), 0) for i in range(24)) minute = dict( (str(i), dict((str(j), 0) for j in range(60))) for i in range(24)) daily = dict((str(i), 0) for i in range(1, 32)) # Perform upserts, setting metadata db.stats.daily.update( { _id: id_daily, hourly: hourly, minute: minute}, { $set: { metadata: daily_metadata }}, upsert=True) db.stats.monthly.update( { _id: id_monthly, daily: daily }, { $set: { m: monthly_metadata }}, upsert=True)
The function pre-allocated both the monthly and daily documents at the same time. The performance benets from separating these operations are negligible, so its reasonable to keep both operations in the same function. Ideally, your application should pre-allocate documents before needing to write data to maintain consistent update performance. Additionally, its important to avoid causing a spike in activity and latency by creating documents all at once. In the following example, document updates (i.e. log_hit()) will also pre-allocate a document probabilistically. However, by tuning probability, you can limit redundant preallocate() calls.
from random import random from datetime import datetime, timedelta, time # Example probability based on 500k hits per day per page prob_preallocate = 1.0 / 500000 def log_hit(db, dt_utc, site, page): if random.random() < prob_preallocate: preallocate(db, dt_utc + timedelta(days=1), site, page) # Update daily stats doc ...
Using this method, there will be a high probability that each document will already exist before your application needs to issue update operations. Youll also be able to prevent a regular spike in activity for pre-allocation, and be able to eliminate document growth.
625
Retrieving Data for a Real-Time Chart This example describes fetching the data from the above MongoDB system, for use in generating a chart that displays the number of hits to a particular resource over the last hour.
Querying
Use the following query in a find_one operation at the Python/PyMongo console to retrieve the number of hits to a specic resource (i.e. http://docs.mongodb.org/manual/index.html) with minute-level granularity:
>>> db.stats.daily.find_one( ... {metadata: {date:dt, site:site-1, page:/index.html}}, ... { minute: 1 })
Use the following query to retrieve the number of hits to a resource over the last day, with hour-level granularity:
>>> db.stats.daily.find_one( ... {metadata: {date:dt, site:site-1, page:/foo.gif}}, ... { hourly: 1 })
If you want a few days of hourly data, you can use a query in the following form:
>>> db.stats.daily.find( ... { ... metadata.date: { $gte: dt1, $lte: dt2 }, ... metadata.site: site-1, ... metadata.page: /index.html}, ... { metadata.date: 1, hourly: 1 } }, ... sort=[(metadata.date, 1)])
Indexing
To support these query operation, create a compound index on the following daily statistics elds: metadata.site, metadata.page, and metadata.date (in that order.) Use the following operation at the Python/PyMongo console.
>>> db.stats.daily.ensure_index([ ... (metadata.site, 1), ... (metadata.page, 1), ... (metadata.date, 1)])
This index makes it possible to efciently run the query for multiple days of hourly data. At the same time, any compound index on page and date, will allow you to query efciently for a single days statistics. Get Data for a Historical Chart
Querying
To retrieve daily data for a single month, use the following query:
>>> db.stats.monthly.find_one( ... {metadata: ... {date:dt, ... site: site-1,
626
... ...
page:/index.html}}, { daily: 1 })
To retrieve several months of daily data, use a variation on the above query:
>>> db.stats.monthly.find( ... { ... metadata.date: { $gte: dt1, $lte: dt2 }, ... metadata.site: site-1, ... metadata.page: /index.html}, ... { metadata.date: 1, daily: 1 } }, ... sort=[(metadata.date, 1)])
Indexing
Create the following index to support these queries for monthly data on the metadata.site, metadata.page, and metadata.date elds:
>>> db.stats.monthly.ensure_index([ ... (metadata.site, 1), ... (metadata.page, 1), ... (metadata.date, 1)])
This eld order will efciently support range queries for a single page over several months.
50.2.4 Sharding
The only potential limits on the performance of this system are the number of shards in your system, and the shard key that you use. An ideal shard key will distribute upserts between the shards while routing all queries to a single shard, or a small number of shards. While your choice of shard key may depend on the precise workload of your deployment, consider using { metadata.site: 1, metadata.page: 1 } as a shard key. The combination of site and page (or event) will lead to a well balanced cluster for most deployments. Enable sharding for the daily statistics collection with the following shardCollection (page 851) command in the Python/PyMongo console:
>>> db.command(shardCollection, stats.daily, { ... key : { metadata.site: 1, metadata.page : 1 } })
Upon success, you will see the following response:

{ "collectionsharded" : "stats.daily", "ok" : 1 }
Enable sharding for the monthly statistics collection with the following shardCollection (page 851) command in the Python/PyMongo console:
>>> db.command(shardCollection, stats.monthly, { ... key : { metadata.site: 1, metadata.page : 1 } })
Upon success, you will see the following response:

{ "collectionsharded" : "stats.monthly", "ok" : 1 }
627
One downside of the { metadata.site: 1, metadata.page: 1 } shard key is: if one page dominates all your trafc, all updates to that page will go to a single shard. This is basically unavoidable, since all update for a single page are going to a single document. You may wish to include the date in addition to the site, and page elds so that MongoDB can split histories so that you can serve different historical ranges with different shards. Use the following shardCollection (page 851) command to shard the daily statistics collection in the Python/PyMongo console:
>>> db.command(shardCollection, stats.daily, { ... key:{metadata.site:1,metadata.page:1,metadata.date:1}}) { "collectionsharded" : "stats.daily", "ok" : 1 }
Enable sharding for the monthly statistics collection with the following shardCollection (page 851) command in the Python/PyMongo console:
>>> db.command(shardCollection, stats.monthly, { ... key:{metadata.site:1,metadata.page:1,metadata.date:1}}) { "collectionsharded" : "stats.monthly", "ok" : 1 }
Note: Determine your actual requirements and load before deciding to shard. In many situations a single MongoDB instance may be able to keep track of all events and pages.
50.3 Hierarchical Aggregation

50.3.1 Overview
Background If you collect a large amount of data, but do not pre-aggregate (page 619), and you want to have access to aggregated information and reports, then you need a method to aggregate these data into a usable form. This document provides an overview of these aggregation patterns and processes. For clarity, this case study assumes that the incoming event data resides in a collection named events. For details on how you might get the event data into the events collection, please see Storing Log Data (page 609) document. This document continues using this example. Solution The rst step in the aggregation process is to aggregate event data into the nest required granularity. Then use this aggregation to generate the next least specic level granularity and this repeat process until you have generated all required views. The solution uses several collections: the raw data (i.e. events) collection as well as collections for aggregated hourly, daily, weekly, monthly, and yearly statistics. All aggregations use the mapReduce (page 814) command, in a hierarchical process. The following gure illustrates the input and output of each job: Note: Aggregating raw events into an hourly collection is qualitatively different from the operation that aggregates hourly statistics into the daily collection. See also: map-reduce, mapReduce (page 814), and the Map-Reduce (page 311) page for more information on the Map-reduce data aggregation paradigm.
628
Figure 50.3: Hierarchy of data aggregation.
50.3.2 Schema
When designing the schema for event storage, its important to track the events included in the aggregation and events that are not yet included. Relational Approach A simple tactic from relational database, uses an auto-incremented integer as the primary key. However, this introduces a signicant performance penalty for event logging process because the aggregation process must fetch new keys one at a time. If you can batch your inserts into the events collection, you can use an auto-increment primary key by using the find_and_modify command to generate the _id values, as in the following example:
>>> obj = db.my_sequence.find_and_modify( ... query={_id:0}, ... update={$inc: {inc: 50}} ... upsert=True, ... new=True) >>> batch_of_ids = range(obj[inc]-50, obj[inc])
However, in most cases you can simply include a timestamp with each event that you can use to distinguish processed events from unprocessed events. This example assumes that you are calculating average session length for logged-in users on a website. The events will have the following form:
{ "userid": "rick", "ts": ISODate(2010-10-10T14:17:22Z), "length":95 }
The operations described in the next session will calculate total and average session times for each user at the hour, day, week, month and year. For each aggregation you will want to store the number of sessions so that MongoDB can incrementally recompute the average session times. The aggregate document will resemble the following: 50.3. Hierarchical Aggregation 629
{ _id: { u: "rick", d: ISODate("2010-10-10T14:00:00Z") }, value: { ts: ISODate(2010-10-10T15:01:00Z), total: 254, count: 10, mean: 25.4 } }
Note: The timestamp value in the _id sub-document, which will allow you to incrementally update documents at various levels of the hierarchy.
50.3.3 Operations
This section assumes that all events exist in the events collection and have a timestamp. The operations, thus are to aggregate from the events collection into the smallest aggregatehourly totals and then aggregate from the hourly totals into coarser granularity levels. In all cases, these operations will store aggregation time as a last_run variable. Creating Hourly Views from Event Collections
Aggregation
Note: Although this solution uses Python and PyMongo to connect with MongoDB, you must pass JavaScript functions (i.e. mapf, reducef, and finalizef) to the mapReduce (page 814) command. Begin by creating a map function, as below:
mapf_hour = bson.Code(function() { var key = { u: this.userid, d: new Date( this.ts.getFullYear(), this.ts.getMonth(), this.ts.getDate(), this.ts.getHours(), 0, 0, 0); emit( key, { total: this.length, count: 1, mean: 0, ts: new Date(); }); })
In this case, it emits key-value pairs that contain the data you want to aggregate as youd expect. The function also emits a ts value that makes it possible to cascade aggregations to coarser grained aggregations (i.e. hour to day, etc.) Consider the following reduce function:
reducef = bson.Code(function(key, values) { var r = { total: 0, count: 0, mean: 0, ts: null }; values.forEach(function(v) { r.total += v.total;
630
r.count += v.count; }); return r; })
The reduce function returns a document in the same format as the output of the map function. This pattern for map and reduce functions makes map-reduce processes easier to test and debug. While the reduce function ignores the mean and ts (timestamp) values, the nalize step, as follows, computes these data:
finalizef = bson.Code(function(key, value) { if(value.count > 0) { value.mean = value.total / value.count; } value.ts = new Date(); return value; })
With the above function the map_reduce operation itself will resemble the following:
cutoff = datetime.utcnow() - timedelta(seconds=60) query = { ts: { $gt: last_run, $lt: cutoff } } db.events.map_reduce( map=mapf_hour, reduce=reducef, finalize=finalizef, query=query, out={ reduce: stats.hourly }) last_run = cutoff
The cutoff variable allows you to process all events that have occurred since the last run but before 1 minute ago. This allows for some delay in logging events. You can safely run this aggregation as often as you like, provided that you update the last_run variable each time.
Indexing
Create an index on the timestamp (i.e. the ts eld) to support the query selection of the map_reduce operation. Use the following operation at the Python/PyMongo console:
>>> db.events.ensure_index(ts)
Deriving Day-Level Data

Aggregation
To calculate daily statistics, use the hourly statistics as input. Begin with the following map function:
mapf_day = bson.Code(function() { var key = { u: this._id.u, d: new Date( this._id.d.getFullYear(), this._id.d.getMonth(),
50.3. Hierarchical Aggregation
631
this._id.d.getDate(), 0, 0, 0, 0) }; emit( key, { total: this.value.total, count: this.value.count, mean: 0, ts: null }); })
The map function for deriving day-level data differs from the initial aggregation above in the following ways: the aggregation key is the (userid, date) rather than (userid, hour) to support daily aggregation. the keys and values emitted (i.e. emit()) are actually the total and count values from the hourly aggregates rather than properties from event documents. This is the case for all the higher-level aggregation operations. Because the output of this map function is the same as the previous map function, you can use the same reduce and nalize functions. The actual code driving this level of aggregation is as follows:
cutoff = datetime.utcnow() - timedelta(seconds=60) query = { value.ts: { $gt: last_run, $lt: cutoff } } db.stats.hourly.map_reduce( map=mapf_day, reduce=reducef, finalize=finalizef, query=query, out={ reduce: stats.daily }) last_run = cutoff
There are a couple of things to note here. First of all, the query is not on ts now, but value.ts, the timestamp written during the nalization of the hourly aggregates. Also note that you are, in fact, aggregating from the stats.hourly collection into the stats.daily collection.
Indexing
Because you will run the query option regularly which nds on the value.ts eld, you may wish to create an index to support this. Use the following operation in the Python/PyMongo shell to create this index:
>>> db.stats.hourly.ensure_index(value.ts)
Weekly and Monthly Aggregation

Aggregation
You can use the aggregated day-level data to generate weekly and monthly statistics. A map function for generating weekly data follows:
632
mapf_week = bson.Code(function() { var key = { u: this._id.u, d: new Date( this._id.d.valueOf() - dt.getDay()*24*60*60*1000) }; emit( key, { total: this.value.total, count: this.value.count, mean: 0, ts: null }); })
Here, to get the group key, the function takes the current and subtracts days until you get the beginning of the week. In the weekly map function, youll use the rst day of the month as the group key, as follows:
mapf_month = bson.Code(function() { d: new Date( this._id.d.getFullYear(), this._id.d.getMonth(), 1, 0, 0, 0, 0) }; emit( key, { total: this.value.total, count: this.value.count, mean: 0, ts: null }); })
These map functions are identical to each other except for the date calculation.
Indexing
Create additional indexes to support the weekly and monthly aggregation options on the value.ts eld. Use the following operation in the Python/PyMongo shell.
>>> db.stats.daily.ensure_index(value.ts) >>> db.stats.monthly.ensure_index(value.ts)
Refactor Map Functions
Use Pythons string interpolation to refactor the map function denitions as follows:
mapf_hierarchical = function() { var key = { u: this._id.u, d: %s }; emit( key, { total: this.value.total, count: this.value.count, mean: 0,
633
ts: null }); } mapf_day = bson.Code( mapf_hierarchical % new Date( this._id.d.getFullYear(), this._id.d.getMonth(), this._id.d.getDate(), 0, 0, 0, 0)) mapf_week = bson.Code( mapf_hierarchical % new Date( this._id.d.valueOf() - dt.getDay()*24*60*60*1000)) mapf_month = bson.Code( mapf_hierarchical % new Date( this._id.d.getFullYear(), this._id.d.getMonth(), 1, 0, 0, 0, 0)) mapf_year = bson.Code( mapf_hierarchical % new Date( this._id.d.getFullYear(), 1, 1, 0, 0, 0, 0))
You can create a h_aggregate function to wrap the map_reduce operation, as below, to reduce code duplication:
def h_aggregate(icollection, ocollection, mapf, cutoff, last_run): query = { value.ts: { $gt: last_run, $lt: cutoff } } icollection.map_reduce( map=mapf, reduce=reducef, finalize=finalizef, query=query, out={ reduce: ocollection.name })
With h_aggregate dened, you can perform all aggregation operations as follows:
cutoff = datetime.utcnow() - timedelta(seconds=60) h_aggregate(db.events, db.stats.hourly, mapf_hour, cutoff, last_run) h_aggregate(db.stats.hourly, db.stats.daily, mapf_day, cutoff, last_run) h_aggregate(db.stats.daily, db.stats.weekly, mapf_week, cutoff, last_run) h_aggregate(db.stats.daily, db.stats.monthly, mapf_month, cutoff, last_run) h_aggregate(db.stats.monthly, db.stats.yearly, mapf_year, cutoff, last_run) last_run = cutoff
As long as you save and restore the last_run variable between aggregations, you can run these aggregations as often as you like since each aggregation operation is incremental.
50.3.4 Sharding
Ensure that you choose a shard key that is not the incoming timestamp, but rather something that varies signicantly in the most recent documents. In the example above, consider using the userid as the most signicant part of the shard key.
634
To prevent a single, active user from creating a large, chunk that MongoDB cannot split, use a compound shard key with (username, timestamp) on the events collection. Consider the following:
>>> db.command(shardCollection,events, { ... key : { userid: 1, ts : 1} } ) { "collectionsharded": "events", "ok" : 1 }
To shard the aggregated collections you must use the _id eld, so you can issue the following group of shard operations in the Python/PyMongo shell:
db.command(shardCollection, key: { _id: 1 } }) db.command(shardCollection, key: { _id: 1 } }) db.command(shardCollection, key: { _id: 1 } }) db.command(shardCollection, key: { _id: 1 } }) stats.daily, { stats.weekly, { stats.monthly, { stats.yearly, {
You should also update the h_aggregate map-reduce wrapper to support sharded output Add sharded:True to the out argument. See the full sharded h_aggregate function:
def h_aggregate(icollection, ocollection, mapf, cutoff, last_run): query = { value.ts: { $gt: last_run, $lt: cutoff } } icollection.map_reduce( map=mapf, reduce=reducef, finalize=finalizef, query=query, out={ reduce: ocollection.name, sharded: True })
635
636
CHAPTER 51
Product Data Management
MongoDBs exible schema makes it particularly well suited to storing information for product data management and e-commerce websites and solutions. The Product Catalog (page 637) document describes methods and practices for modeling and managing a product catalog using MongoDB, while the Inventory Management (page 645) document introduces a pattern for handling interactions between inventory and users shopping carts. Finally the Category Hierarchy (page 651) document describes methods for interacting with category hierarchies in MongoDB.
51.1 Product Catalog

51.1.1 Overview
This document describes the basic patterns and principles for designing an E-Commerce product catalog system using MongoDB as a storage engine. Problem Product catalogs must have the capacity to store many differed types of objects with different sets of attributes. These kinds of data collections are quite compatible with MongoDBs data model, but many important considerations and design decisions remain. Solution For relational databases, there are several solutions that address this problem, each with a different performance prole. This section examines several of these options and then describes the preferred MongoDB solution. SQL and Relational Data Models
Concrete Table Inheritance
One approach, in a relational model, is to create a table for each product category. Consider the following example SQL statement for creating database tables:
CREATE TABLE product_audio_album ( sku char(8) NOT NULL, ... artist varchar(255) DEFAULT NULL,
637
genre_0 varchar(255) DEFAULT NULL, genre_1 varchar(255) DEFAULT NULL, ..., PRIMARY KEY(sku)) ... CREATE TABLE product_film ( sku char(8) NOT NULL, ... title varchar(255) DEFAULT NULL, rating char(8) DEFAULT NULL, ..., PRIMARY KEY(sku)) ...
This approach has limited exibility for two key reasons: You must create a new table for every new category of products. You must explicitly tailor all queries for the exact type of product.
Single Table Inheritance
Another relational data model uses a single table for all product categories and adds new columns anytime you need to store data regarding a new type of product. Consider the following SQL statement:
CREATE TABLE product ( sku char(8) NOT NULL, ... artist varchar(255) DEFAULT NULL, genre_0 varchar(255) DEFAULT NULL, genre_1 varchar(255) DEFAULT NULL, ... title varchar(255) DEFAULT NULL, rating char(8) DEFAULT NULL, ..., PRIMARY KEY(sku))
This approach is more exible than concrete table inheritance: it allows single queries to span different product types, but at the expense of space.
Multiple Table Inheritance
Also in the relational model, you may use a multiple table inheritance pattern to represent common attributes in a generic product table, with some variations in individual category product tables. Consider the following SQL statement:
CREATE TABLE product ( sku char(8) NOT NULL, title varchar(255) DEFAULT NULL, description varchar(255) DEFAULT NULL, price, ... PRIMARY KEY(sku)) CREATE TABLE product_audio_album ( sku char(8) NOT NULL, ... artist varchar(255) DEFAULT NULL,
638
Chapter 51. Product Data Management
genre_0 varchar(255) DEFAULT NULL, genre_1 varchar(255) DEFAULT NULL, ..., PRIMARY KEY(sku), FOREIGN KEY(sku) REFERENCES product(sku)) ... CREATE TABLE product_film ( sku char(8) NOT NULL, ... title varchar(255) DEFAULT NULL, rating char(8) DEFAULT NULL, ..., PRIMARY KEY(sku), FOREIGN KEY(sku) REFERENCES product(sku)) ...
Multiple table inheritance is more space-efcient than single table inheritance (page 638) and somewhat more exible than concrete table inheritance (page 638). However, this model does require an expensive JOIN operation to obtain all relevant attributes relevant to a product.
Entity Attribute Values
The nal substantive pattern from relational modeling is the entity-attribute-value schema where you would create a meta-model for product data. In this approach, you maintain a table with three columns, e.g. entity_id, attribute_id, value, and these triples describe each product. Consider the description of an audio recording. You may have a series of rows representing the following relationships: Entity sku_00e8da9b sku_00e8da9b sku_00e8da9b sku_00e8da9b sku_00e8da9b sku_00e8da9b ... Attribute type title ... artist genre genre ... Value Audio Album A Love Supreme ... John Coltrane Jazz General ...
This schema is totally exible: any entity can have any set of any attributes. New product categories do not require any changes to the data model in the database. The downside for these models, is that all nontrivial queries require large numbers of JOIN operations that results in large performance penalties.
Avoid Modeling Product Data
Additionally some e-commerce solutions with relational database systems avoid choosing one of the data models above, and serialize all of this data into a BLOB column. While simple, the details become difcult to access for search and sort.
51.1. Product Catalog
639
Non-Relational Data Model
Because MongoDB is a non-relational database, the data model for your product catalog can benet from this additional exibility. The best models use a single MongoDB collection to store all the product data, which is similar to the single table inheritance (page 638) relational model. MongoDBs dynamic schema means that each document need not conform to the same schema. As a result, the document for each product only needs to contain attributes relevant to that product. Schema At the beginning of the document, the schema must contain general product information, to facilitate searches of the entire catalog. Then, a details sub-document that contains elds that vary between product types. Consider the following example document for an album product.
{ sku: "00e8da9b", type: "Audio Album", title: "A Love Supreme", description: "by John Coltrane", asin: "B0000A118M", shipping: { weight: 6, dimensions: { width: 10, height: 10, depth: 1 }, }, pricing: { list: 1200, retail: 1100, savings: 100, pct_savings: 8 }, details: { title: "A Love Supreme [Original Recording Reissued]", artist: "John Coltrane", genre: [ "Jazz", "General" ], ... tracks: [ "A Love Supreme Part I: Acknowledgement", "A Love Supreme Part II - Resolution", "A Love Supreme, Part III: Pursuance", "A Love Supreme, Part IV-Psalm" ], }, }
A movie item would have the same elds for general product information, shipping, and pricing, but have different details sub-document. Consider the following:
{ sku: "00e8da9d", type: "Film",
640
..., asin: "B000P0J0AQ", shipping: { ... }, pricing: { ... }, details: { title: "The Matrix", director: [ "Andy Wachowski", "Larry Wachowski" ], writer: [ "Andy Wachowski", "Larry Wachowski" ], ..., aspect_ratio: "1.66:1" }, }
Note: In MongoDB, you can have elds that hold multiple values (i.e. arrays) without any restrictions on the number of elds or values (as with genre_0 and genre_1) and also without the need for a JOIN operation.
51.1.2 Operations
For most deployments the primary use of the product catalog is to perform search operations. This section provides an overview of various types of queries that may be useful for supporting an e-commerce site. All examples in this document use the Python programming language and the PyMongo driver for MongoDB, but you can implement this system using any language you choose. Find Albums by Genre and Sort by Year Produced
Querying
This query returns the documents for the products of a specic genre, sorted in reverse chronological order:
query = db.products.find({type:Audio Album, details.genre: jazz}) query = query.sort([(details.issue_date, -1)])
Indexing
To support this query, create a compound index on all the properties used in the lter and in the sort:
db.products.ensure_index([ (type, 1), (details.genre, 1), (details.issue_date, -1)])
Note: The nal component of the index is the sort eld. This allows MongoDB to traverse the index in the sorted order to preclude a slow in-memory sort.
641
Find Products Sorted by Percentage Discount Descending While most searches will be for a particular type of product (e.g album, movie, etc.,) in some situations you may want to return all products in a certain price range, or discount percentage.
Querying
To return this data use the pricing information that exists in all products to nd the products with the highest percentage discount:
query = db.products.find( { pricing.pct_savings: {$gt: 25 }) query = query.sort([(pricing.pct_savings, -1)])
Indexing
To support this type of query, you will want to create an index on the pricing.pct_savings eld:
db.products.ensure_index(pricing.pct_savings)
Since MongoDB can read indexes in ascending or descending order, the order of the index does not matter. Note: If you want to preform range queries (e.g. return all products over $25) and then sort by another property like pricing.retail, MongoDB cannot use the index as effectively in this situation. The eld that you want to select a range, or perform sort operations, must be the last eld in a compound index in order to avoid scanning an entire collection. Using different properties within a single combined range query and sort operation requires some scanning which will limit the speed of your query.
Find Movies Based on Staring Actor

Querying
Use the following query to select documents within the details of a specied product type (i.e. Film) of product (a movie) to nd products that contain a certain value (i.e. a specic actor in the details.actor eld,) with the results sorted by date descending:
query = db.products.find({type: Film, details.actor: Keanu Reeves}) query = query.sort([(details.issue_date, -1)])
Indexing
To support this query, you may want to create the following index.
db.products.ensure_index([ (type, 1), (details.actor, 1), (details.issue_date, -1)])
This index begins with the type eld and then narrows by the other search eld, where the nal component of the index is the sort eld to maximize index efciency.
642
Find Movies with a Particular Word in the Title Regardless of database engine, in order to retrieve this information the system will need to scan some number of documents or records to satisfy this query.
Querying
MongoDB supports regular expressions within queries. In Python, you can use the python:re module to construct the query:
import re re_hacker = re.compile(r.*hacker.*, re.IGNORECASE) query = db.products.find({type: Film, title: re_hacker}) query = query.sort([(details.issue_date, -1)])
MongoDB provides a special syntax for regular expression queries without the need for the re module. Consider the following alternative which is equivalent to the above example:
query = db.products.find({ type: Film, title: {$regex: .*hacker.*, $options:i}}) query = query.sort([(details.issue_date, -1)])
The $options (page 774) operator species a case insensitive match.

Indexing
The indexing strategy for these kinds of queries is different from previous attempts. Here, create an index on { type: 1, details.issue_date: -1, title: 1 } using the following command at the Python/PyMongo console:
db.products.ensure_index([ (type, 1), (details.issue_date, -1), (title, 1)])
This index makes it possible to avoid scanning whole documents by using the index for scanning the title rather than forcing MongoDB to scan whole documents for the title eld. Additionally, to support the sort on the details.issue_date eld, by placing this eld before the title eld, ensures that the result set is already ordered before MongoDB lters title eld.
51.1.3 Scaling
Sharding Database performance for these kinds of deployments are dependent on indexes. You may use sharding to enhance performance by allowing MongoDB to keep larger portions of those indexes in RAM. In sharded congurations, select a shard key that allows mongos (page 1032) to route queries directly to a single shard or small group of shards. Since most of the queries in this system include the type eld, include this in the shard key. Beyond this, the remainder of the shard key is difcult to predict without information about your databases actual activity and distribution. Consider that:
643
details.issue_date would be a poor addition to the shard key because, although it appears in a number of queries, no queries were selective by this eld. you should include one or more elds in the detail document that you query frequently, and a eld that has quasi-random features, to prevent large unsplittable chunks. In the following example, assume that the details.genre eld is the second-most queried eld after type. Enable sharding using the following shardCollection (page 851) operation at the Python/PyMongo console:
>>> db.command(shardCollection, product, { ... key : { type: 1, details.genre : 1, sku:1 } }) { "collectionsharded" : "details.genre", "ok" : 1 }
Note: Even if you choose a poor shard key that requires mongos (page 1032) to broadcast all to all shards, you will still see some benets from sharding, because: 1. Sharding makes a larger amount of memory available to store indexes, and 2. MongoDB will parallelize queries across shards, reducing latency.
Read Preference While sharding is the best way to scale operations, some data sets make it impossible to partition data so that mongos (page 1032) can route queries to specic shards. In these situations mongos (page 1032) sends the query to all shards and then combines the results before returning to the client. In these situations, you can add additional read performance by allowing mongos (page 1032) to read from the secondary instances in a replica set by conguring read preference in your client. Read preference is congurable on a per-connection or per-operation basis. In PyMongo, set the read_preference argument. The SECONDARY property in the following example, permits reads from a secondary (as well as a primary) for the entire connection .
conn = pymongo.MongoClient(read_preference=pymongo.SECONDARY)
Conversely, the SECONDARY_ONLY read preference means that the client will only send read operation only to the secondary member
conn = pymongo.MongoClient(read_preference=pymongo.SECONDARY_ONLY)
You can also specify read_preference for specic queries, as follows:

results = db.product.find(..., read_preference=pymongo.SECONDARY)
or
results = db.product.find(..., read_preference=pymongo.SECONDARY_ONLY)
See also: Replica Set Read Preference (page 402)
644
51.2 Inventory Management

51.2.1 Overview
This case study provides an overview of practices and patterns for designing and developing the inventory management portions of an E-commerce application. See also: Product Catalog (page 637). Problem Customers in e-commerce stores regularly add and remove items from their shopping cart, change quantities multiple times, abandon the cart at any point, and sometimes have problems during and after checkout that require a hold or canceled order. These activities make it difcult to maintain inventory systems and counts and ensure that customers cannot buy items that are unavailable while they shop in your store. Solution This solution keeps the traditional metaphor of the shopping cart, but the shopping cart will age. After a shopping cart has been inactive for a certain period of time, all items in the cart re-enter the available inventory and the cart is empty. The state transition diagram for a shopping cart is below:
Schema Inventory collections must maintain counts of the current available inventory of each stock-keeping unit (SKU; or item) as well as a list of items in carts that may return to the available inventory if they are in a shopping cart that times out. In the following example, the _id eld stores the SKU:
{ _id: 00e8da9b, qty: 16, carted: [ { qty: 1, cart_id: 42, timestamp: ISODate("2012-03-09T20:55:36Z"), }, { qty: 2, cart_id: 43,
51.2. Inventory Management
645
timestamp: ISODate("2012-03-09T21:55:36Z"), }, ] }
Note: These examples use a simplied schema. In a production implementation, you may choose to merge this schema with the product catalog schema described in the Product Catalog (page 637) document. The SKU above has 16 items in stock, 1 item a cart, and 2 items in a second cart. This leaves a total of 19 unsold items of merchandise. To model the shopping cart objects, you need to maintain sku, quantity, elds embedded in a shopping cart document:
{ _id: 42, last_modified: ISODate("2012-03-09T20:55:36Z"), status: active, items: [ { sku: 00e8da9b, qty: 1, item_details: {...} }, { sku: 0ab42f88, qty: 4, item_details: {...} } ] }
Note: The item_details eld in each line item allows your application to display the cart contents to the user without requiring a second query to fetch details from the catalog collection.
51.2.2 Operations
This section introduces operations that you may use to support an e-commerce site. All examples in this document use the Python programming language and the PyMongo driver for MongoDB, but you can implement this system using any language you choose. Add an Item to a Shopping Cart Moving an item from the available inventory to a cart is a fundamental requirement for a shopping cart system. The most important requirement is to ensure that your application will never move an unavailable item from the inventory to the cart. Ensure that inventory is only updated if there is sufcient inventory to satisfy the request with the following add_item_to_cart function operation.
def add_item_to_cart(cart_id, sku, qty, details): now = datetime.utcnow() # Make sure the cart is still active and add the line item result = db.cart.update( {_id: cart_id, status: active }, { $set: { last_modified: now }, $push: { items: {sku: sku, qty:qty, details: details } } }, w=1) if not result[updatedExisting]: raise CartInactive()
646
# Update the inventory result = db.inventory.update( {_id:sku, qty: {$gte: qty}}, {$inc: {qty: -qty}, $push: { carted: { qty: qty, cart_id:cart_id, timestamp: now } } }, w=1) if not result[updatedExisting]: # Roll back our cart update db.cart.update( {_id: cart_id }, { $pull: { items: {sku: sku } } }) raise InadequateInventory()
The system does not trust that the available inventory can satisfy a request First this operation checks to make sure that the cart is active before adding an item. Then, it veries that the available inventory to satisfy the request before decrementing inventory. If there is not adequate inventory, the system removes the cart update: by specifying w=1 and checking the result allows the application to report an error if the cart is inactive or available quantity is insufcient to satisfy the request. Note: This operation requires no indexes beyond the default index on the _id eld.
Modifying the Quantity in the Cart The following process underlies adjusting the quantity of items in a users cart. The application must ensure that when a user increases the quantity of an item, in addition to updating the carted entry for the users cart, that the inventory exists to cover the modication.
def update_quantity(cart_id, sku, old_qty, new_qty): now = datetime.utcnow() delta_qty = new_qty - old_qty # Make sure the cart is still active and add the line item result = db.cart.update( {_id: cart_id, status: active, items.sku: sku }, {$set: { last_modified: now, items.$.qty: new_qty }, }, w=1) if not result[updatedExisting]: raise CartInactive() # Update the inventory result = db.inventory.update( {_id:sku, carted.cart_id: cart_id, qty: {$gte: delta_qty} }, {$inc: {qty: -delta_qty }, $set: { carted.$.qty: new_qty, timestamp: now } }, w=1) if not result[updatedExisting]: # Roll back our cart update
647
db.cart.update( {_id: cart_id, items.sku: sku }, {$set: { items.$.qty: old_qty } }) raise InadequateInventory()
Note: That the positional operator $ updates the particular carted entry and item that matched the query. This allows the application to update the inventory and keep track of the data needed to rollback the cart in a single atomic operation. The code also ensures that the cart is active. Note: This operation requires no indexes beyond the default index on the _id eld.
Checking Out The checkout operation must: validate the method of payment and remove the carted items after the transaction succeeds. Consider the following procedure:
def checkout(cart_id): now = datetime.utcnow() # Make sure the cart is still active and set to pending. Also # fetch the cart details so we can calculate the checkout price cart = db.cart.find_and_modify( {_id: cart_id, status: active }, update={$set: { status: pending,last_modified: now } } ) if cart is None: raise CartInactive() # Validate payment details; collect payment try: collect_payment(cart) db.cart.update( {_id: cart_id }, {$set: { status: complete } } ) db.inventory.update( {carted.cart_id: cart_id}, {$pull: {cart_id: cart_id} }, multi=True) except: db.cart.update( {_id: cart_id }, {$set: { status: active } } ) raise
Begin by locking the cart by setting its status to pending Then the system will verify that the cart is still active and collect payment data. Then, the findAndModify (page 826) command makes it possible to update the cart atomically and return its details to capture payment information. Then: If the payment is successful, then the application will remove the carted items from the inventory documents and set the cart to complete. If payment is unsuccessful, the application will unlock the cart by setting its status to active and report a payment error. Note: This operation requires no indexes beyond the default index on the _id eld.
648
Returning Inventory from Timed-Out Carts

Process
Periodically, your application must expire inactive carts and return their items to available inventory. In the example that follows the variable timeout controls the length of time before a cart expires:
def expire_carts(timeout): now = datetime.utcnow() threshold = now - timedelta(seconds=timeout) # Lock and find all the expiring carts db.cart.update( {status: active, last_modified: { $lt: threshold } }, {$set: { status: expiring } }, multi=True ) # Actually expire each cart for cart in db.cart.find({status: expiring}): # Return all line items to inventory for item in cart[items]: db.inventory.update( { _id: item[sku], carted.cart_id: cart[id], carted.qty: item[qty] }, {$inc: { qty: item[qty] }, $pull: { carted: { cart_id: cart[id] } } }) db.cart.update( {_id: cart[id] }, {$set: { status: expired })
This procedure: 1. nds all carts that are older than the threshold and are due for expiration. 2. for each expiring cart, return all items to the available inventory. 3. once the items return to the available inventory, set the status eld to expired.
Indexing
To support returning inventory from timed-out cart, create an index to support queries on their status and last_modified elds. Use the following operations in the Python/PyMongo shell:
db.cart.ensure_index([(status, 1), (last_modified, 1)])
Error Handling The above operations do not account for one possible failure situation: if an exception occurs after updating the shopping cart but before updating the inventory collection. This would result in a shopping cart that may be absent or expired but items have not returned to available inventory.
649
To account for this case, your application will need a periodic cleanup operation that nds inventory items that have carted items and check that to ensure that they exist in a users cart, and return them to available inventory if they do not.
def cleanup_inventory(timeout): now = datetime.utcnow() threshold = now - timedelta(seconds=timeout) # Find all the expiring carted items for item in db.inventory.find( {carted.timestamp: {$lt: threshold }}): # Find all the carted items that matched carted = dict( (carted_item[cart_id], carted_item) for carted_item in item[carted] if carted_item[timestamp] < threshold) # First Pass: Find any carts that are active and refresh the carted items for cart in db.cart.find( { _id: {$in: carted.keys() }, status:active}): cart = carted[cart[_id]] db.inventory.update( { _id: item[_id], carted.cart_id: cart[_id] }, { $set: {carted.$.timestamp: now } }) del carted[cart[_id]] # Second Pass: All the carted items left in the dict need to now be # returned to inventory for cart_id, carted_item in carted.items(): db.inventory.update( { _id: item[_id], carted.cart_id: cart_id, carted.qty: carted_item[qty] }, { $inc: { qty: carted_item[qty] }, $pull: { carted: { cart_id: cart_id } } })
To summarize: This operation nds all carted items that have time stamps older than the threshold. Then, the process makes two passes over these items: 1. Of the items with time stamps older than the threshold, if the cart is still active, it resets the time stamp to maintain the carts. 2. Of the stale items that remain in inactive carts, the operation returns these items to the inventory. Note: The function above is safe for use because it checks to ensure that the cart has expired before returning items from the cart to inventory. However, it could be long-running and slow other updates and queries. Use judiciously.
51.2.3 Sharding
If you need to shard the data for this system, the _id eld is an ideal shard key for both carts and products because most update operations use the _id eld. This allows mongos (page 1032) to route all updates that select on _id to
650
a single mongod (page 1021) process. There are two drawbacks for using _id as a shard key: If the cart collections _id is an incrementing value, all new carts end up on a single shard. You can mitigate this effect by choosing a random value upon the creation of a cart, such as a hash (i.e. MD5 or SHA-1) of an ObjectID, as the _id. The process for this operation would resemble the following:
import hashlib import bson cart_id = bson.ObjectId() cart_id_hash = hashlib.md5(str(cart_id)).hexdigest() cart = { "_id": cart_id, "cart_hash": cart_id_hash } db.cart.insert(cart)
Cart expiration and inventory adjustment requires update operations and queries to broadcast to all shards when using _id as a shard key. This may be less relevant as the expiration functions run relatively infrequently and you can queue them or articially slow them down (as with judicious use of sleep()) to minimize server load. Use the following commands in the Python/PyMongo console to shard the cart and inventory collections:
>>> db.command(shardCollection, inventory ... key: { _id: 1 } ) { "collectionsharded" : "inventory", "ok" : 1 } >>> db.command(shardCollection, cart) ... key: { _id: 1 } ) { "collectionsharded" : "cart", "ok" : 1 }
51.3 Category Hierarchy

51.3.1 Overview
This document provides the basic design for modeling a product hierarchy stored in MongoDB as well as a collection of common operations for interacting with this data that will help you begin to write an E-commerce product category hierarchy. See also: Product Catalog (page 637) Solution To model a product category hierarchy, this solution keeps each category in its own document that also has a list of its ancestors or parents. This document uses music genres as the basis of its examples: Because these kinds of categories change infrequently, this model focuses on the operations needed to keep the hierarchy up-to-date rather than the performance prole of update operations. Schema This schema has the following properties:
51.3. Category Hierarchy
651
Figure 51.1: Initial category hierarchy A single document represents each category in the hierarchy. An ObjectId identies each category document for internal cross-referencing. Each category document has a human-readable name and a URL compatible slug eld. The schema stores a list of ancestors for each category to facilitate displaying a query and its ancestors using only a single query. Consider the following prototype:
{ "_id" : ObjectId("4f5ec858eb03303a11000002"), "name" : "Modal Jazz", "parent" : ObjectId("4f5ec858eb03303a11000001"), "slug" : "modal-jazz", "ancestors" : [ { "_id" : ObjectId("4f5ec858eb03303a11000001"), "slug" : "bop", "name" : "Bop" }, { "_id" : ObjectId("4f5ec858eb03303a11000000"), "slug" : "ragtime", "name" : "Ragtime" } ] }
51.3.2 Operations
This section outlines the category hierarchy manipulations that you may need in an E-Commerce site. All examples in this document use the Python programming language and the PyMongo driver for MongoDB, but you can implement this system using any language you choose.
652
Read and Display a Category

Querying
Use the following option to read and display a category hierarchy. This query will use the slug eld to return the category information and a bread crumb trail from the current category to the top level category.
category = db.categories.find( {slug:slug}, {_id:0, name:1, ancestors.slug:1, ancestors.name:1 })
Indexing
Create a unique index on the slug eld with the following operation on the Python/PyMongo console:
>>> db.categories.ensure_index(slug, unique=True)
Add a Category to the Hierarchy To add a category you must rst determine its ancestors. Take adding a new category Swing as a child of Ragtime, as below:
Figure 51.2: Adding a category The insert operation would be trivial except for the ancestors. To dene this array, consider the following helper function:
def build_ancestors(_id, parent_id): parent = db.categories.find_one( {_id: parent_id},
653
{name: 1, slug: 1, ancestors:1}) parent_ancestors = parent.pop(ancestors) ancestors = [ parent ] + parent_ancestors db.categories.update( {_id: _id}, {$set: { ancestors: ancestors } })
You only need to travel up one level in the hierarchy to get the ancestor list for Ragtime that you can use to build the ancestor list for Swing. Then create a document with the following set of operations:
doc = dict(name=Swing, slug=swing, parent=ragtime_id) swing_id = db.categories.insert(doc) build_ancestors(swing_id, ragtime_id)
Note: Since these queries and updates all selected based on _id, you only need the default MongoDB-supplied index on _id to support this operation efciently.
Change the Ancestry of a Category This section address the process for reorganizing the hierarchy by moving bop under swing as follows:
Figure 51.3: Change the parent of a category
654
Procedure
Update the bop document to reect the change in ancestry with the following operation:
db.categories.update( {_id:bop_id}, {$set: { parent: swing_id } } )
The following helper function, rebuilds the ancestor elds to ensure correctness.
def build_ancestors_full(_id, parent_id): ancestors = [] while parent_id is not None: parent = db.categories.find_one( {_id: parent_id}, {parent: 1, name: 1, slug: 1, ancestors:1}) parent_id = parent.pop(parent) ancestors.append(parent) db.categories.update( {_id: _id}, {$set: { ancestors: ancestors } })
You can use the following loop to reconstruct all the descendants of the bop category:
for cat in db.categories.find( {ancestors._id: bop_id}, {parent_id: 1}): build_ancestors_full(cat[_id], cat[parent_id])
Indexing
Create an index on the ancestors._id eld to support the update operation.

db.categories.ensure_index(ancestors._id)
Rename a Category To a rename a category you need to both update the category itself and also update all the descendants. Consider renaming Bop to BeBop as in the following gure: First, you need to update the category name with the following operation:
db.categories.update( {_id:bop_id}, {$set: { name: BeBop } } )
Next, you need to update each descendants ancestors list:

db.categories.update( {ancestors._id: bop_id}, {$set: { ancestors.$.name: BeBop } }, multi=True)
This operation uses: the positional operation $ to match the exact ancestor entry that matches the query, and
1 Your application cannot guarantee that the ancestor list of a parent category is correct, because MongoDB may process the categories out-oforder.
655
Figure 51.4: Rename a category
656
the multi option to update all documents that match this query. Note: In this case, the index you have already dened on ancestors._id is sufcient to ensure good performance.
51.3.3 Sharding
For most deployments, sharding this collection has limited value because the collection will be very small. If you do need to shard, because most updates query the _id eld, this eld is a suitable shard key. Shard the collection with the following operation in the Python/PyMongo console.
>>> db.command(shardCollection, categories, { ... key: {_id: 1} }) { "collectionsharded" : "categories", "ok" : 1 }
657
658
CHAPTER 52
Content Management Systems
The content management use cases introduce fundamental MongoDB practices and approaches, using familiar problems and simple examples. The Metadata and Asset Management (page 659) document introduces a model that you may use when designing a web site content management system, while Storing Comments (page 666) introduces the method for modeling user comments on content, like blog posts, and media, in MongoDB.
52.1 Metadata and Asset Management

52.1.1 Overview
This document describes the design and pattern of a content management system using MongoDB modeled on the popular Drupal CMS. Problem You are designing a content management system (CMS) and you want to use MongoDB to store the content of your sites. Solution To build this system you will use MongoDBs exible schema to store all content nodes in a single collection regardless of type. This guide will provide prototype schema and describe common operations for the following primary node types: Basic Page Basic pages are useful for displaying infrequently-changing text such as an about page. With a basic page, the salient information is the title and the content. Blog entry Blog entries record a stream of posts from users on the CMS and store title, author, content, and date as relevant information. Photo Photos participate in photo galleries, and store title, description, author, and date along with the actual photo binary data. This solution does not describe schema or process for storing or using navigational and organizational information.
659
Schema Although documents in the nodes collection contain content of different types, all documents have a similar structure and a set of common elds. Consider the following prototype document for a basic page node type:
{ _id: ObjectId(...), nonce: ObjectId(...), metadata: { type: basic-page section: my-photos, slug: about, title: About Us, created: ISODate(...), author: { _id: ObjectId(...), name: Rick }, tags: [ ... ], detail: { text: # About Us\n... } } }
Most elds are descriptively titled. The section eld identies groupings of items, as in a photo gallery, or a particular blog . The slug eld holds a URL-friendly unique representation of the node, usually that is unique within its section for generating URLs. All documents also have a detail eld that varies with the document type. For the basic page above, the detail eld might hold the text of the page. For a blog entry, the detail eld might hold a sub-document. Consider the following prototype:
{ ... metadata: { ... type: blog-entry, section: my-blog, slug: 2012-03-noticed-the-news, ... detail: { publish_on: ISODate(...), text: I noticed the news from Washington today... } } }
Photos require a different approach. Because photos can be potentially larger than these documents, its important to separate the binary photo storage from the nodes metadata. GridFS provides the ability to store larger les in MongoDB. GridFS stores data in two collections, in this case, cms.assets.files, which stores metadata, and cms.assets.chunks which stores the data itself. Consider the following prototype document from the cms.assets.files collection:
{ _id: ObjectId(...), length: 123..., chunkSize: 262144, uploadDate: ISODate(...), contentType: image/jpeg, md5: ba49a..., metadata: { nonce: ObjectId(...),
660
Chapter 52. Content Management Systems
slug: 2012-03-invisible-bicycle, type: photo, section: my-album, title: Kitteh, created: ISODate(...), author: { _id: ObjectId(...), name: Jared }, tags: [ ... ], detail: { filename: kitteh_invisible_bike.jpg, resolution: [ 1600, 1600 ], ... } } }
Note: This document embeds the basic node document elds, which allows you to use the same code to manipulate nodes, regardless of type.
52.1.2 Operations
This section outlines a number of common operations for building and interacting with the metadata and asset layer of the cms for all node types. All examples in this document use the Python programming language and the PyMongo driver for MongoDB, but you can implement this system using any language you choose. Create and Edit Content Nodes
Procedure
The most common operations inside of a CMS center on creating and editing content. Consider the following insert() operation:
db.cms.nodes.insert({ nonce: ObjectId(), metadata: { section: myblog, slug: 2012-03-noticed-the-news, type: blog-entry, title: Noticed in the News, created: datetime.utcnow(), author: { id: user_id, name: Rick }, tags: [ news, musings ], detail: { publish_on: datetime.utcnow(), text: I noticed the news from Washington today... } } })
Once inserted, your application must have some way of preventing multiple concurrent updates. The schema uses the special nonce eld to help detect concurrent edits. By using the nonce eld in the query portion of the update operation, the application will generate an error if there is an editing collision. Consider the following update
def update_text(section, slug, nonce, text): result = db.cms.nodes.update( { metadata.section: section, metadata.slug: slug, nonce: nonce },
52.1. Metadata and Asset Management
661
{ $set:{metadata.detail.text: text, nonce: ObjectId() } }, w=1) if not result[updatedExisting]: raise ConflictError()
You may also want to perform metadata edits to the item such as adding tags:
db.cms.nodes.update( { metadata.section: section, metadata.slug: slug }, { $addToSet: { tags: { $each: [ interesting, funny ] } } })
In this example the $addToSet (page 789) operator will only add values to the tags eld if they do not already exist in the tags array, theres no need to supply or update the nonce.
Index Support
To support updates and queries on the metadata.section, and metadata.slug, elds and to ensure that two editors dont create two documents with the same section name or slug. Use the following operation at the Python/PyMongo console:
>>> db.cms.nodes.ensure_index([ ... (metadata.section, 1), (metadata.slug, 1)], unique=True)
The unique=True option prevents to documents from colliding. If you want an index to support queries on the above elds and the nonce eld create the following index:
>>> db.cms.nodes.ensure_index([ ... (metadata.section, 1), (metadata.slug, 1), (nonce, 1) ])
However, in most cases, the rst index will be sufcient to support these operations. Upload a Photo
Procedure
To update a photo object, use the following operation, which builds upon the basic update procedure:
def upload_new_photo( input_file, section, slug, title, author, tags, details): fs = GridFS(db, cms.assets) with fs.new_file( content_type=image/jpeg, metadata=dict( type=photo, locked=datetime.utcnow(), section=section, slug=slug, title=title, created=datetime.utcnow(), author=author, tags=tags, detail=detail)) as upload_file: while True: chunk = input_file.read(upload_file.chunk_size) if not chunk: break upload_file.write(chunk)
662
# unlock the file db.assets.files.update( {_id: upload_file._id}, {$set: { locked: None } } )
Because uploading the photo spans multiple documents and is a non-atomic operation, you must lock the le during upload by writing datetime.utcnow() in the record. This helps when there are multiple concurrent editors and lets the application detect stalled le uploads. This operation assumes that, for photo upload, the last update will succeed:
def update_photo_content(input_file, section, slug): fs = GridFS(db, cms.assets) # Delete the old version if its unlocked or was locked more than 5 # minutes ago file_obj = db.cms.assets.find_one( { metadata.section: section, metadata.slug: slug, metadata.locked: None }) if file_obj is None: threshold = datetime.utcnow() - timedelta(seconds=300) file_obj = db.cms.assets.find_one( { metadata.section: section, metadata.slug: slug, metadata.locked: { $lt: threshold } }) if file_obj is None: raise FileDoesNotExist() fs.delete(file_obj[_id]) # update content, keep metadata unchanged file_obj[locked] = datetime.utcnow() with fs.new_file(**file_obj): while True: chunk = input_file.read(upload_file.chunk_size) if not chunk: break upload_file.write(chunk) # unlock the file db.assets.files.update( {_id: upload_file._id}, {$set: { locked: None } } )
As with the basic operations, you can use a much more simple operation to edit the tags:
db.cms.assets.files.update( { metadata.section: section, metadata.slug: slug }, { $addToSet: { metadata.tags: { $each: [ interesting, funny ] } } })
Index Support
Create a unique index on { metadata.section: 1, metadata.slug: 1 } to support the above operations and prevent users from creating or updating the same le concurrently. Use the following operation in the Python/PyMongo console:
>>> db.cms.assets.files.ensure_index([ ... (metadata.section, 1), (metadata.slug, 1)], unique=True)
663
Locate and Render a Node To locate a node based on the value of metadata.section and metadata.slug, use the following find_one operation.
node = db.nodes.find_one({metadata.section: section, metadata.slug: slug })
Note: The index dened (section, slug) created to support the update operation, is sufcient to support this operation as well.
Locate and Render a Photo To locate an image based on the value of metadata.section and metadata.slug, use the following find_one operation.
fs = GridFS(db, cms.assets) with fs.get_version({metadata.section: section, metadata.slug: slug }) as img_fpo: # do something with the image file
Note: The index dened (section, slug) created to support the update operation, is sufcient to support this operation as well.
Search for Nodes by Tag

Querying
To retrieve a list of nodes based on their tags, use the following query:
nodes = db.nodes.find({metadata.tags: tag })
Indexing
Create an index on the tags eld in the cms.nodes collection, to support this query:
>>> db.cms.nodes.ensure_index(tags)
Search for Images by Tag

Procedure
To retrieve a list of images based on their tags, use the following operation:
image_file_objects = db.cms.assets.files.find({metadata.tags: tag }) fs = GridFS(db, cms.assets) for image_file_object in db.cms.assets.files.find( {metadata.tags: tag }): image_file = fs.get(image_file_object[_id]) # do something with the image file
664
Indexing
Create an index on the tags eld in the cms.assets.files collection, to support this query:
>>> db.cms.assets.files.ensure_index(tags)
Generate a Feed of Recently Published Blog Articles

Querying
Use the following operation to generate a list of recent blog posts sorted in descending order by date, for use on the index page of your site, or in an .rss or .atom feed.
articles = db.nodes.find({ metadata.section: my-blog metadata.published: { $lt: datetime.utcnow() } }) articles = articles.sort({metadata.published: -1})
Note: In many cases you will want to limit the number of nodes returned by this query.
Indexing
Create a compound index on the { metadata.section: port this query and sort operation.
1, metadata.published:
1 } elds to sup-
>>> db.cms.nodes.ensure_index( ... [ (metadata.section, 1), (metadata.published, -1) ])
Note: For all sort or range queries, ensure that eld with the sort or range operation is the nal eld in the index.
52.1.3 Sharding
In a CMS, read performance is more critical than write performance. To achieve the best read performance in a sharded cluster, ensure that the mongos (page 1032) can route queries to specic shards. Also remember that MongoDB can not enforce unique indexes across shards. Using a compound shard key that consists of metadata.section and metadata.slug, will provide the same semantics as describe above. Warning: Consider the actual use and workload of your cluster before conguring sharding for your cluster. Use the following operation at the Python/PyMongo shell:
>>> db.command(shardCollection, cms.nodes, { ... key : { metadata.section: 1, metadata.slug : 1 } }) { "collectionsharded": "cms.nodes", "ok": 1} >>> db.command(shardCollection, cms.assets.files, { ... key : { metadata.section: 1, metadata.slug : 1 } }) { "collectionsharded": "cms.assets.files", "ok": 1}
To shard the cms.assets.chunks collection, you must use the _id eld as the shard key. The following operation will shard the collection
665
>>> db.command(shardCollection, cms.assets.chunks, { ... key : { files_id: 1 } }) { "collectionsharded": "cms.assets.chunks", "ok": 1}
Sharding on the files_id eld ensures routable queries because all reads from GridFS must rst look up the document in cms.assets.files and then look up the chunks separately.
52.2 Storing Comments

This document outlines the basic patterns for storing user-submitted comments in a content management system (CMS.)
52.2.1 Overview
MongoDB provides a number of different approaches for storing data like users-comments on content from a CMS. There is no correct implementation, but there are a number of common approaches and known considerations for each approach. This case study explores the implementation details and trade offs of each option. The three basic patterns are: 1. Store each comment in its own document. This approach provides the greatest exibility at the expense of some additional application level complexity. These implementations make it possible to display comments in chronological or threaded order, and place no restrictions on the number of comments attached to a specic object. 2. Embed all comments in the parent document. This approach provides the greatest possible performance for displaying comments at the expense of exibility: the structure of the comments in the document controls the display format. Note: Because of the limit on document size (page 1105), documents, including the original content and all comments, cannot grow beyond 16 megabytes. 3. A hybrid design, stores comments separately from the parent, but aggregates comments into a small number of documents, where each contains many comments. Also consider that comments can be threaded, where comments are always replies to parent item or to another comment, which carries certain architectural requirements discussed below.
52.2.2 One Document per Comment

Schema If you store each comment in its own document, the documents in your comments collection, would have the following structure:
{ _id: ObjectId(...), discussion_id: ObjectId(...), slug: 34db, posted: ISODateTime(...), author: { id: ObjectId(...),
666
name: Rick }, text: This is so bogus ... }
This form is only suitable for displaying comments in chronological order. Comments store: the discussion_id eld that references the discussion parent, a URL-compatible slug identier, a posted timestamp, an author sub-document that contains a reference to a users prole in the id eld and their name in the name eld, and the full text of the comment. To support threaded comments, you might use a slightly different structure like the following:
{ _id: ObjectId(...), discussion_id: ObjectId(...), parent_id: ObjectId(...), slug: 34db/8bda full_slug: 2012.02.08.12.21.08:34db/2012.02.09.22.19.16:8bda, posted: ISODateTime(...), author: { id: ObjectId(...), name: Rick }, text: This is so bogus ... }
This structure: adds a parent_id eld that stores the contents of the _id eld of the parent comment, modies the slug eld to hold a path composed of the parent or parents slug and this comments unique slug, and adds a full_slug eld that that combines the slugs and time information to make it easier to sort documents in a threaded discussion by date. Warning: MongoDB can only index 1024 bytes (page 1106). This includes all eld data, the eld name, and the namespace (i.e. database name and collection name.) This may become an issue when you create an index of the full_slug eld to support sorting.
Operations This section contains an overview of common operations for interacting with comments represented using a schema where each comment is its own document. All examples in this document use the Python programming language and the PyMongo driver for MongoDB, but you can implement this system using any language you choose. Issue the following commands at the interactive Python shell to load the required libraries:
>>> import bson >>> import pymongo
52.2. Storing Comments
667
Post a New Comment
To post a new comment in a chronologically ordered (i.e. without threading) system, use the following insert() operation:
slug = generate_pseudorandom_slug() db.comments.insert({ discussion_id: discussion_id, slug: slug, posted: datetime.utcnow(), author: author_info, text: comment_text })
To insert a comment for a system with threaded comments, you must generate the slug path and full_slug at insert. See the following operation:
posted = datetime.utcnow() # generate the unique portions of the slug and full_slug slug_part = generate_pseudorandom_slug() full_slug_part = posted.strftime(%Y.%m.%d .%H.%M.%S) + : + slug_part # load the parent comment (if any) if parent_slug: parent = db.comments.find_one( {discussion_id: discussion_id, slug: parent_slug }) slug = parent[slug] + / + slug_part full_slug = parent[full_slug] + / + full_slug_part else: slug = slug_part full_slug = full_slug_part # actually insert the comment db.comments.insert({ discussion_id: discussion_id, slug: slug, full_slug: full_slug, posted: posted, author: author_info, text: comment_text })
View Paginated Comments
To view comments that are not threaded, select all comments participating in a discussion and sort by the posted eld. For example:
cursor cursor cursor cursor = = = = db.comments.find({discussion_id: discussion_id}) cursor.sort(posted) cursor.skip(page_num * page_size) cursor.limit(page_size)
Because the full_slug eld contains both hierarchical information (via the path) and chronological information, you can use a simple sort on the full_slug eld to retrieve a threaded view:
cursor cursor cursor cursor = = = = db.comments.find({discussion_id: discussion_id}) cursor.sort(full_slug) cursor.skip(page_num * page_size) cursor.limit(page_size)
668
See also: cursor.limit (page 959), cursor.skip (page 964), and cursor.sort (page 965)
Indexing
To support the above queries efciently, maintain two compound indexes, on: 1. (discussion_id, posted) and 2. (discussion_id, full_slug) Issue the following operation at the interactive Python shell.
>>> db.comments.ensure_index([ ... (discussion_id, 1), (posted, 1)]) >>> db.comments.ensure_index([ ... (discussion_id, 1), (full_slug, 1)])
Note: Ensure that you always sort by the nal element in a compound index to maximize the performance of these queries.
Retrieve Comments via Direct Links

Queries
To directly retrieve a comment, without needing to page through all comments, you can select by the slug eld:
comment = db.comments.find_one({ discussion_id: discussion_id, slug: comment_slug})
You can retrieve a sub-discussion, or a comment and all of its descendants recursively, by performing a regular expression prex query on the full_slug eld:
import re subdiscussion = db.comments.find_one({ discussion_id: discussion_id, full_slug: re.compile(^ + re.escape(parent_slug)) }) subdiscussion = subdiscussion.sort(full_slug)
Indexing
Since you have already created indexes on { discussion_id: 1, full_slug: } to support retrieving sub-discussions, you can add support for the above queries by adding an index on { discussion_id: 1 , slug: 1 }. Use the following operation in the Python shell:
>>> db.comments.ensure_index([ ... (discussion_id, 1), (slug, 1)])
669
52.2.3 Embedding All Comments

This design embeds the entire discussion of a comment thread inside of the topic document. In this example, the topic, document holds the total content for whatever content youre managing. Schema Consider the following prototype topic document:
{ _id: ObjectId(...), ... lots of topic data ... comments: [ { posted: ISODateTime(...), author: { id: ObjectId(...), name: Rick }, text: This is so bogus ... }, ... ] }
This structure is only suitable for a chronological display of all comments because it embeds comments in chronological order. Each document in the array in the comments contains the comments date, author, and text. Note: Since youre storing the comments in sorted order, there is no need to maintain per-comment slugs. To support threading using this design, you would need to embed comments within comments, using a structure that resembles the following:
{ _id: ObjectId(...), ... lots of topic data ... replies: [ { posted: ISODateTime(...), author: { id: ObjectId(...), name: Rick }, text: This is so bogus ... , replies: [ { author: { ... }, ... }, ... ] }
Here, the replies eld in each comment holds the sub-comments, which can intern hold sub-comments. Note: In the embedded document design, you give up some exibility regarding display format, because it is difcult to display comments except as you store them in MongoDB. If, in the future, you want to switch from chronological to threaded or from threaded to chronological, this design would make that migration quite expensive. Warning: Remember that BSON documents have a 16 megabyte size limit (page 1105). If popular discussions grow larger than 16 megabytes, additional document growth will fail. Additionally, when MongoDB documents grow signicantly after creation you will experience greater storage fragmentation and degraded update performance while MongoDB migrates documents internally.
670
Operations This section contains an overview of common operations for interacting with comments represented using a schema that embeds all comments the document of the parent or topic content. Note: For all operations below, there is no need for any new indexes since all the operations are function within documents. Because you would retrieve these documents by the _id eld, you can rely on the index that MongoDB creates automatically.
Post a new comment
To post a new comment in a chronologically ordered (i.e unthreaded) system, you need the following update():
db.discussion.update( { discussion_id: discussion_id }, { $push: { comments: { posted: datetime.utcnow(), author: author_info, text: comment_text } } } )
The $push (page 792) operator inserts comments into the comments array in correct chronological order. For threaded discussions, the update() operation is more complex. To reply to a comment, the following code assumes that it can retrieve the path as a list of positions, for the parent comment:
if path != []: str_path = ..join(replies.%d % part for part in path) str_path += .replies else: str_path = replies db.discussion.update( { discussion_id: discussion_id }, { $push: { str_path: { posted: datetime.utcnow(), author: author_info, text: comment_text } } } )
This constructs a eld name of the form replies.0.replies.2... as str_path and then uses this value with the $push (page 792) operator to insert the new comment into the parent comments replies array.
To view the comments in a non-threaded design, you must use the $slice (page 793) operator:
discussion = db.discussion.find_one( {discussion_id: discussion_id}, { ... some fields relevant to your page from the root discussion ..., comments: { $slice: [ page_num * page_size, page_size ] } })
To return paginated comments for the threaded design, you must retrieve the whole document and paginate the comments within the application:
discussion = db.discussion.find_one({discussion_id: discussion_id})
671
def iter_comments(obj): for reply in obj[replies]: yield reply for subreply in iter_comments(reply): yield subreply paginated_comments = itertools.slice( iter_comments(discussion), page_size * page_num, page_size * (page_num + 1))
Retrieve a Comment via Direct Links
Instead of retrieving comments via slugs as above, the following example retrieves comments using their position in the comment list or tree. For chronological (i.e. non-threaded) comments, just use the $slice (page 793) operator to extract a comment, as follows:
discussion = db.discussion.find_one( {discussion_id: discussion_id}, {comments: { $slice: [ position, position ] } }) comment = discussion[comments][0]
For threaded comments, you must nd the correct path through the tree in your application, as follows:
discussion = db.discussion.find_one({discussion_id: discussion_id}) current = discussion for part in path: current = current.replies[part] comment = current
Note: Since parent comments embed child replies, this operation actually retrieves the entire sub-discussion for the comment you queried for. See also: find_one().
52.2.4 Hybrid Schema Design

Schema In the hybrid approach you will store comments in buckets that hold about 100 comments. Consider the following example document:
{ _id: ObjectId(...), discussion_id: ObjectId(...), page: 1, count: 42, comments: [ { slug: 34db, posted: ISODateTime(...), author: { id: ObjectId(...), name: Rick },
672
text: This is so bogus ... }, ... ] }
Each document maintains page and count data that contains meta data regarding the page, the page number and the comment count, in addition to the comments array that holds the comments themselves. Note: Using a hybrid format makes storing threaded comments complex, and this specic conguration is not covered in this document. Also, 100 comments is a soft limit for the number of comments per page. This value is arbitrary: choose a value that will prevent the maximum document size from growing beyond the 16MB BSON document size limit (page 1105), but large enough to ensure that most comment threads will t in a single document. In some situations the number of comments per document can exceed 100, but this does not affect the correctness of the pattern.
Operations This section contains a number of common operations that you may use when building a CMS using this hybrid storage model with documents that hold 100 comment pages. All examples in this document use the Python programming language and the PyMongo driver for MongoDB, but you can implement this system using any language you choose.
Post a New Comment
Updating In order to post a new comment, you need to $push (page 792) the comment onto the last page and $inc (page 784) that pages comment count. Consider the following example that queries on the basis of a discussion_id eld:
page = db.comment_pages.find_and_modify( { discussion_id: discussion[_id], page: discussion[num_pages] }, { $inc: { count: 1 }, $push: { comments: { slug: slug, ... } } }, fields={count:1}, upsert=True, new=True )
The find_and_modify() operation is an upsert,: if MongoDB cannot nd a document with the correct page number, the find_and_modify() will create it and initialize the new document with appropriate values for count and comments. To limit the number of comments per page to roughly 100, you will need to create new pages as they become necessary. Add the following logic to support this:
if page[count] > 100: db.discussion.update( { discussion_id: discussion[_id], num_pages: discussion[num_pages] }, { $inc: { num_pages: 1 } } )
This update() operation includes the last known number of pages in the query to prevent a race condition where the number of pages increments twice, that would result in a nearly or totally empty document. If another process increments the number of pages, then update above does nothing.
673
Indexing To support the find_and_modify() and update() operations, maintain a compound index on (discussion_id, page) in the comment_pages collection, by issuing the following operation at the Python/PyMongo console:
>>> db.comment_pages.ensure_index([ ... (discussion_id, 1), (page, 1)])
The following function denes how to paginate comments with a xed page size (i.e. not with the roughly 100 comment documents in the above example,) as en example:
def find_comments(discussion_id, skip, limit): result = [] page_query = db.comment_pages.find( { discussion_id: discussion_id }, { count: 1, comments: { $slice: [ skip, limit ] } }) page_query = page_query.sort(page) for page in page_query: result += page[comments] skip = max(0, skip - page[count]) limit -= len(page[comments]) if limit == 0: break return result
Here, the $slice (page 793) operator pulls out comments from each page, but only when this satises the skip requirement. For example: if you have 3 pages with 100, 102, 101, and 22 comments on each page, and you wish to retrieve comments where skip=300 and limit=50. Use the following algorithm: Skip Limit Discussion 300 50 {$slice: [ 300, 50 ] } matches nothing in page #1; subtract page #1s count from skip and continue. 200 50 {$slice: [ 200, 50 ] } matches nothing in page #2; subtract page #2s count from skip and continue. 98 50 {$slice: [ 98, 50 ] } matches 2 comments in page #3; subtract page #3s count from skip (saturating at 0), subtract 2 from limit, and continue. 0 48 {$slice: [ 0, 48 ] } matches all 22 comments in page #4; subtract 22 from limit and continue. 0 26 There are no more pages; terminate loop. Note: Since you already have an index on (discussion_id, page) in your comment_pages collection, MongoDB can satisfy these queries efciently.
Retrieve a Comment via Direct Links
Query To retrieve a comment directly without paging through all preceding pages of commentary, use the slug to nd the correct page, and then use application logic to nd the correct comment:
page = db.comment_pages.find_one( { discussion_id: discussion_id, comments.slug: comment_slug}, { comments: 1 }) for comment in page[comments]: if comment[slug] = comment_slug: break
674
Indexing To perform this query efciently youll need a new index on the discussion_id and comments.slug elds (i.e. { discussion_id: 1 comments.slug: 1 }.) Create this index using the following operation in the Python/PyMongo console:
>>> db.comment_pages.ensure_index([ ... (discussion_id, 1), (comments.slug, 1)])
52.2.5 Sharding
For all of the architectures discussed above, you will want to the discussion_id eld to participate in the shard key, if you need to shard your application. For applications that use the one document per comment approach, consider using slug (or full_slug, in the case of threaded comments) elds in the shard key to allow the mongos (page 1032) instances to route requests by slug. Issue the following operation at the Python/PyMongo console:
>>> db.command(shardCollection, comments, { ... key : { discussion_id : 1, full_slug: 1 } })
This will return the following response:

{ "collectionsharded" : "comments", "ok" : 1 }
In the case of comments that fully-embedded in parent content documents the determination of the shard key is outside of the scope of this document. For hybrid documents, use the page number of the comment page in the shard key along with the discussion_id to allow MongoDB to split popular discussions between, while grouping discussions on the same shard. Issue the following operation at the Python/PyMongo console:
>>> db.command(shardCollection, comment_pages, { ... key : { discussion_id : 1, page: 1 } }) { "collectionsharded" : "comment_pages", "ok" : 1 }
675
676
CHAPTER 53
Python Application Development
53.1 Write a Tumblelog Application with Django MongoDB Engine

53.1.1 Introduction
In this tutorial, you will learn how to create a basic tumblelog application using the popular Django Python webframework and the MongoDB database. The tumblelog will consist of two parts: 1. A public site that lets people view posts and comment on them. 2. An admin site that lets you add, change and delete posts and publish comments. This tutorial assumes that you are already familiar with Django and have a basic familiarity with MongoDB operation and have installed MongoDB (page 3). Where to get help If youre having trouble going through this tutorial, please post a message to mongodb-user or join the IRC chat in #mongodb on irc.freenode.net to chat with other MongoDB users who might be able to help. Note: Django MongoDB Engine uses a forked version of Django 1.3 that adds non-relational support.
53.1.2 Installation
Begin by installing packages required by later steps in this tutorial. Prerequisite This tutorial uses pip to install packages and virtualenv to isolate Python environments. While these tools and this conguration are not required as such, they ensure a standard environment and are strongly recommended. Issue the following commands at the system prompt:
pip install virtualenv virtualenv myproject
Respectively, these commands: install the virtualenv program (using pip) and create an isolated Python environment for this project (named myproject.)
677
To activate myproject environment at the system prompt, use the following command:
source myproject/bin/activate
Installing Packages Django MongoDB Engine directly depends on: Django-nonrel, a fork of Django 1.3 that adds support for non-relational databases djangotoolbox, a bunch of utilities for non-relational Django applications and backends Install by issuing the following commands:
pip install https://bitbucket.org/wkornewald/django-nonrel/get/tip.tar.gz pip install https://bitbucket.org/wkornewald/djangotoolbox/get/tip.tar.gz pip install https://github.com/django-nonrel/mongodb-engine/tarball/master
Continue with the tutorial to begin building the tumblelog application.
53.1.3 Build a Blog to Get Started

In this tutorial you will build a basic blog as the foundation of this application and use this as the basis of your tumblelog application. You will add the rst post using the shell and then later use the Django administrative interface. Call the startproject command, as with other Django projects, to get started and create the basic project skeleton:
django-admin.py startproject tumblelog
Conguring Django Congure the database in the tumblelog/settings.py le:

DATABASES = { default: { ENGINE: django_mongodb_engine, NAME: my_tumble_log } }
See also: The Django MongoDB Engine Settings documentation for more conguration options. Dene the Schema The rst step in writing a tumblelog in Django is to dene the models or in MongoDBs terminology documents. In this application, you will dene posts and comments, so that each Post can contain a list of Comments. Edit the tumblelog/models.py le so it resembles the following:
from django.db import models from django.core.urlresolvers import reverse from djangotoolbox.fields import ListField, EmbeddedModelField
678
Chapter 53. Python Application Development
class Post(models.Model): created_at = models.DateTimeField(auto_now_add=True, db_index=True) title = models.CharField(max_length=255) slug = models.SlugField() body = models.TextField() comments = ListField(EmbeddedModelField(Comment), editable=False) def get_absolute_url(self): return reverse(post, kwargs={"slug": self.slug}) def __unicode__(self): return self.title class Meta: ordering = ["-created_at"]
class Comment(models.Model): created_at = models.DateTimeField(auto_now_add=True) body = models.TextField(verbose_name="Comment") author = models.CharField(verbose_name="Name", max_length=255)
The Django nonrel code looks the same as vanilla Django, however there is no built in support for some of MongoDBs native data types like Lists and Embedded data. djangotoolbox handles these denitions. See also: The Django MongoDB Engine elds documentation for more. The models declare an index to the Post class. One for the created_at date as our frontpage will order by date: there is no need to add db_index on SlugField because there is a default index on SlugField. Add Data with the Shell The manage.py provides a shell interface for the application that you can use to insert data into the tumblelog. Begin by issuing the following command to load the Python shell:
python manage.py shell
Create the rst post using the following sequence of operations:

>>> >>> ... ... ... ... >>> from tumblelog.models import * post = Post( title="Hello World!", slug="hello-world", body = "Welcome to my new shiny Tumble log powered by MongoDB and Django-MongoDB!" ) post.save()
Add comments using the following sequence of operations:

>>> [] >>> ... ... >>> >>> post.comments comment = Comment( author="Joe Bloggs", body="Great post! Im looking forward to reading your blog") post.comments.append(comment) post.save()
53.1. Write a Tumblelog Application with Django MongoDB Engine
679
Finally, inspect the post:

>>> post = Post.objects.get() >>> post <Post: Hello World!> >>> post.comments [<Comment: Comment object>]
Add the Views Because django-mongodb provides tight integration with Django you can use generic views to display the frontpage and post pages for the tumblelog. Insert the following content into the urls.py le to add the views:
from django.conf.urls.defaults import patterns, include, url from django.views.generic import ListView, DetailView from tumblelog.models import Post urlpatterns = patterns(, url(r^$, ListView.as_view( queryset=Post.objects.all(), context_object_name="posts_list"), name="home" ), url(r^post/(?P<slug>[a-zA-Z0-9-]+)/$, DetailView.as_view( queryset=Post.objects.all(), context_object_name="post"), name="post" ), )
Add Templates In the tumblelog directory add the following directories templates and templates/tumblelog for storing the tumblelog templates:
mkdir -p templates/tumblelog
Congure Django so it can nd the templates by updating TEMPLATE_DIRS in the settings.py le to the following:
import os.path TEMPLATE_DIRS = ( os.path.join(os.path.realpath(__file__), ../templates), )
Then add a base template that all others can inherit from. Add the following to templates/base.html:
<!DOCTYPE html> <html lang="en"> <head> <meta charset="utf-8"> <title>My Tumblelog</title> <link href="http://twitter.github.com/bootstrap/1.4.0/bootstrap.css" rel="stylesheet"> <style>.content {padding-top: 80px;}</style> </head> <body>
680
<div class="topbar"> <div class="fill"> <div class="container"> <h1><a href="/" class="brand">My Tumblelog</a>! <small>Starring MongoDB and Django-MongoDB. </div> </div> </div> <div class="container"> <div class="content"> {% block page_header %}{% endblock %} {% block content %}{% endblock %} </div> </div> </body> </html>
Create the frontpage for the blog, which should list all the posts. templates/tumblelog/post_list.html:
{% extends "base.html" %}
Add the following template to the
{% block content %} {% for post in posts_list %} <h2><a href="{% url post slug=post.slug %}">{{ post.title }}</a></h2> <p>{{ post.body|truncatewords:20 }}</p> <p> {{ post.created_at }} | {% with total=post.comments|length %} {{ total }} comment{{ total|pluralize }} {% endwith %} </p> {% endfor %} {% endblock %}
Finally, add templates/tumblelog/post_detail.html for the individual posts:

{% extends "base.html" %} {% block page_header %} <div class="page-header"> <h1>{{ post.title }}</h1> </div> {% endblock %} {% block content %} <p>{{ post.body }}<p> <p>{{ post.created_at }}</p> <hr> <h2>Comments</h2> {% if post.comments %} {% for comment in post.comments %} <p>{{ comment.body }}</p> <p><strong>{{ comment.author }}</strong> <small>on {{ comment.created_at }}</small></p> {{ comment.text }} {% endfor %} {% endif %}
681
{% endblock %}
Run python manage.py runserver to see your new tumblelog! Go to http://localhost:8000/ and you should see:
53.1.4 Add Comments to the Blog

In the next step you will provide the facility for readers of the tumblelog to comment on posts. This a requires custom form and view to handle the form, and data. You will also update the template to include the form. Create the Comments Form You must customize form handling to deal with embedded comments. By extending ModelForm, it is possible to append the comment to the post on save. Create and add the following to forms.py:
from django.forms import ModelForm from tumblelog.models import Comment
class CommentForm(ModelForm): def __init__(self, object, *args, **kwargs): """Override the default to store the original document that comments are embedded in. """ self.object = object return super(CommentForm, self).__init__(*args, **kwargs) def save(self, *args): """Append to the comments list and save the post""" self.object.comments.append(self.instance) self.object.save() return self.object class Meta: model = Comment
Handle Comments in the View You must extend the generic views need to handle the form logic. Add the following to the views.py le:
682
from django.http import HttpResponseRedirect from django.views.generic import DetailView from tumblelog.forms import CommentForm
class PostDetailView(DetailView): methods = [get, post] def get(self, request, *args, **kwargs): self.object = self.get_object() form = CommentForm(object=self.object) context = self.get_context_data(object=self.object, form=form) return self.render_to_response(context) def post(self, request, *args, **kwargs): self.object = self.get_object() form = CommentForm(object=self.object, data=request.POST) if form.is_valid(): form.save() return HttpResponseRedirect(self.object.get_absolute_url()) context = self.get_context_data(object=self.object, form=form) return self.render_to_response(context)
Note: The PostDetailView class extends the DetailView class so that it can handle GET and POST requests. On POST, post() validates the comment: if valid, post() appends the comment to the post. Dont forget to update the urls.py le and import the PostDetailView class to replace the DetailView class. Add Comments to the Templates Finally, you can add the form to the templates, so that readers can create comments. Splitting the template for the forms out into templates/_forms.html will allow maximum reuse of forms code:
<fieldset> {% for field in form.visible_fields %} <div class="clearfix {% if field.errors %}error{% endif %}"> {{ field.label_tag }} <div class="input"> {{ field }} {% if field.errors or field.help_text %} <span class="help-inline"> {% if field.errors %} {{ field.errors|join: }} {% else %} {{ field.help_text }} {% endif %} </span> {% endif %} </div> </div> {% endfor %} {% csrf_token %} <div style="display:none">{% for h in form.hidden_fields %} {{ h }}{% endfor %}</div> </fieldset>
683
After the comments section in post_detail.html add the following code to generate the comments form:
<h2>Add a comment</h2> <form action="." method="post"> {% include "_forms.html" %} <div class="actions"> <input type="submit" class="btn primary" value="comment"> </div> </form>
Your tumblelogs readers can now comment on your posts! Run python manage.py runserver to see the changes by visiting http://localhost:8000/hello-world/. You should see the following:
684
53.1.5 Add Site Administration Interface

While you may always add posts using the shell interface as above, you can easily create an administrative interface for posts with Django. Enable the admin by adding the following apps to INSTALLED_APPS in settings.py. django.contrib.admin django_mongodb_engine djangotoolbox 53.1. Write a Tumblelog Application with Django MongoDB Engine 685
tumblelog Warning: This application does not require the Sites framework. As a result, remove django.contrib.sites from INSTALLED_APPS. If you need it later please read SITE_ID issues document. Create a admin.py le and register the Post model with the admin app:
from django.contrib import admin from tumblelog.models import Post admin.site.register(Post)
Note: The above modications deviate from the default django-nonrel and djangotoolbox mode of operation. Djangos administration module will not work unless you exclude the comments eld. By making the comments eld non-editable in the admin model denition, you will allow the administrative interface to function. If you need an administrative interface for a ListField you must write your own Form / Widget. See also: The Django Admin documentation docs for additional information. Update the urls.py to enable the administrative interface. Add the import and discovery mechanism to the top of the le and then add the admin import rule to the urlpatterns:
# Enable admin from django.contrib import admin admin.autodiscover() urlpatterns = patterns(, # ... url(râdmin/, include(admin.site.urls)), )
Finally, add a superuser and setup the indexes by issuing the following command at the system prompt:
python manage.py syncdb
Once done run the server and you can login to admin by going to http://localhost:8000/admin/.
686
53.1.6 Convert the Blog to a Tumblelog

Currently, the application only supports posts. In this section you will add special post types including: Video, Image and Quote to provide a more traditional tumblelog application. Adding this data requires no migration. In models.py update the Post class to add new elds for the new post types. Mark these elds with blank=True so that the elds can be empty. Update Post in the models.py les to resemble the following:
POST_CHOICES = ( (p, post), (v, video), (i, image), (q, quote), )
class Post(models.Model): created_at = models.DateTimeField(auto_now_add=True) title = models.CharField(max_length=255) slug = models.SlugField() comments = ListField(EmbeddedModelField(Comment), editable=False) post_type = models.CharField(max_length=1, choices=POST_CHOICES, default=p) body = models.TextField(blank=True, help_text="The body of the Post / Quote") embed_code = models.TextField(blank=True, help_text="The embed code for video") image_url = models.URLField(blank=True, help_text="Image src") author = models.CharField(blank=True, max_length=255, help_text="Author name")
687
def get_absolute_url(self): return reverse(post, kwargs={"slug": self.slug}) def __unicode__(self): return self.title
Note: Django-Nonrel doesnt support multi-table inheritance. This means that you will have to manually create an administrative form to handle data validation for the different post types. The Abstract Inheritance facility means that the view logic would need to merge data from multiple collections. The administrative interface should now handle adding multiple types of post. To conclude this process, you must update the frontend display to handle and output the different post types. In the post_list.html le, change the post output display to resemble the following:
{% if post.post_type == p %} <p>{{ post.body|truncatewords:20 }}</p> {% endif %} {% if post.post_type == v %} {{ post.embed_code|safe }} {% endif %} {% if post.post_type == i %} <p><img src="{{ post.image_url }}" /><p> {% endif %} {% if post.post_type == q %} <blockquote>{{ post.body|truncatewords:20 }}</blockquote> <p>{{ post.author }}</p> {% endif %}
In the post_detail.html le, change the output for full posts:

{% if post.post_type == p %} <p>{{ post.body }}<p> {% endif %} {% if post.post_type == v %} {{ post.embed_code|safe }} {% endif %} {% if post.post_type == i %} <p><img src="{{ post.image_url }}" /><p> {% endif %} {% if post.post_type == q %} <blockquote>{{ post.body }}</blockquote> <p>{{ post.author }}</p> {% endif %}
Now you have a fully edged tumbleblog using Django and MongoDB!
688
53.2 Write a Tumblelog Application with Flask and MongoEngine

53.2.1 Introduction
This tutorial describes the process for creating a basic tumblelog application using the popular Flask Python webframework in conjunction with the MongoDB database. The tumblelog will consist of two parts: 53.2. Write a Tumblelog Application with Flask and MongoEngine 689
1. A public site that lets people view posts and comment on them. 2. An admin site that lets you add and change posts. This tutorial assumes that you are already familiar with Flask and have a basic familiarity with MongoDB and have installed MongoDB (page 3). This tutorial uses MongoEngine as the Object Document Mapper (ODM,) this component may simplify the interaction between Flask and MongoDB. Where to get help If youre having trouble going through this tutorial, please post a message to mongodb-user or join the IRC chat in #mongodb on irc.freenode.net to chat with other MongoDB users who might be able to help.
53.2.2 Installation
Begin by installing packages required by later steps in this tutorial. Prerequisite This tutorial uses pip to install packages and virtualenv to isolate Python environments. While these tools and this conguration are not required as such, they ensure a standard environment and are strongly recommended. Issue the following command at the system prompt:
pip install virtualenv virtualenv myproject
Respectively, these commands: install the virtualenv program (using pip) and create an isolated Python environment for this project (named myproject.) To activate myproject environment at the system prompt, use the following command:
source myproject/bin/activate
Install Packages Flask is a microframework, because it provides a small core of functionality and is highly extensible. For the tumblelog project, this tutorial includes task and the following extension: WTForms provides easy form handling. Flask-MongoEngine provides integration between MongoEngine, Flask, and WTForms. Flask-Script for an easy to use development server Install with the following commands:
pip pip pip pip pip install install install install install flask flask-script WTForms mongoengine flask_mongoengine
Continue with the tutorial to begin building the tumblelog application.
690
53.2.3 Build a Blog to Get Started

First, create a simple bare bones application. Make a directory named tumblelog for the project and then, add the following content into a le named __init__.py:
from flask import Flask app = Flask(__name__)
if __name__ == __main__: app.run()
Next, create the manage.py le. a development server and shell:
Use this le to load additional Flask-scripts in the future. Flask-scripts provides
# Set the path import os, sys sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), ..))) from flask.ext.script import Manager, Server from tumblelog import app manager = Manager(app) # Turn on debugger by default and reloader manager.add_command("runserver", Server( use_debugger = True, use_reloader = True, host = 0.0.0.0) ) if __name__ == "__main__": manager.run()
You can run this application with a test server, by issuing the following command at the system prompt:
python manage.py runserver
There should be no errors, and you can visit http://localhost:5000/ in a web browser to view a page with a 404 message. Congure MongoEngine and Flask Install the Flask extension and add the conguration. Update tumblelog/__init__.py so that it resembles the following:
from flask import Flask from flask.ext.mongoengine import MongoEngine app = Flask(__name__) app.config["MONGODB_SETTINGS"] = {DB: "my_tumble_log"} app.config["SECRET_KEY"] = "KeepThisS3cr3t" db = MongoEngine(app) if __name__ == __main__: app.run()
1
This concept will be familiar to users of Django.
53.2. Write a Tumblelog Application with Flask and MongoEngine
691
See also: The MongoEngine Settings documentation for additional conguration options. Dene the Schema The rst step in writing a tumblelog in Flask is to dene the models or in MongoDBs terminology documents. In this application, you will dene posts and comments, so that each Post can contain a list of Comments. Edit the models.py le so that it resembles the following:
import datetime from flask import url_for from tumblelog import db
class Post(db.Document): created_at = db.DateTimeField(default=datetime.datetime.now, required=True) title = db.StringField(max_length=255, required=True) slug = db.StringField(max_length=255, required=True) body = db.StringField(required=True) comments = db.ListField(db.EmbeddedDocumentField(Comment)) def get_absolute_url(self): return url_for(post, kwargs={"slug": self.slug}) def __unicode__(self): return self.title meta = { allow_inheritance: True, indexes: [-created_at, slug], ordering: [-created_at] }
class Comment(db.EmbeddedDocument): created_at = db.DateTimeField(default=datetime.datetime.now, required=True) body = db.StringField(verbose_name="Comment", required=True) author = db.StringField(verbose_name="Name", max_length=255, required=True)
As above, MongoEngine syntax is simple and declarative. If you have a Django background, the syntax may look familiar. This example denes indexes for Post: one for the created_at date as our frontpage will order by date and another for the individual post slug. Add Data with the Shell The manage.py provides a shell interface for the application that you can use to insert data into the tumblelog. Before conguring the urls and views for this application, you can use this interface to interact with your the tumblelog. Begin by issuing the following command to load the Python shell:
python manage.py shell
Create the rst post using the following sequence of operations:

>>> from tumblelog.models import * >>> post = Post(
692
... ... ... ... >>>
title="Hello World!", slug="hello-world", body="Welcome to my new shiny Tumble log powered by MongoDB, MongoEngine, and Flask" ) post.save()
Add comments using the following sequence of operations:

>>> [] >>> ... ... ... >>> >>> post.comments comment = Comment( author="Joe Bloggs", body="Great post! Im looking forward to reading your blog!" ) post.comments.append(comment) post.save()
Finally, inspect the post:

>>> post = Post.objects.get() >>> post <Post: Hello World!> >>> post.comments [<Comment: Comment object>]
Add the Views Using Flasks class-based views system allows you to produce List and Detail views for tumblelog posts. Add views.py and create a posts blueprint:
from flask import Blueprint, request, redirect, render_template, url_for from flask.views import MethodView from tumblelog.models import Post, Comment posts = Blueprint(posts, __name__, template_folder=templates)
class ListView(MethodView): def get(self): posts = Post.objects.all() return render_template(posts/list.html, posts=posts)
class DetailView(MethodView): def get(self, slug): post = Post.objects.get_or_404(slug=slug) return render_template(posts/detail.html, post=post)
# Register the urls posts.add_url_rule(/, view_func=ListView.as_view(list)) posts.add_url_rule(/<slug>/, view_func=DetailView.as_view(detail))
Now in __init__.py register the blueprint, avoiding a circular dependency by registering the blueprints in a method. Add the following code to the module:
693
def register_blueprints(app): # Prevents circular imports from tumblelog.views import posts app.register_blueprint(posts) register_blueprints(app)
Add this method and method call to the main body of the module and not in the main block. Add Templates In the tumblelog directory add the templates and templates/posts directories to store the tumblelog templates:
mkdir -p templates/posts
Create a base template. All other templates will inherit from this template, which should exist in the templates/base.html le:
<!DOCTYPE html> <html lang="en"> <head> <meta charset="utf-8"> <title>My Tumblelog</title> <link href="http://twitter.github.com/bootstrap/1.4.0/bootstrap.css" rel="stylesheet"> <style>.content {padding-top: 80px;}</style> </head> <body>
{%- block topbar -%} <div class="topbar"> <div class="fill"> <div class="container"> <h2> <a href="/" class="brand">My Tumblelog</a> <small>Starring Flask, MongoDB and MongoEngi </h2> </div> </div> </div> {%- endblock -%} <div class="container"> <div class="content"> {% block page_header %}{% endblock %} {% block content %}{% endblock %} </div> </div> {% block js_footer %}{% endblock %} </body> </html>
Continue by creating a landing page for the blog that will list all posts. templates/posts/list.html le:
{% extends "base.html" %} {% block content %}
Add the following to the
694
{% for post in posts %} <h2><a href="{{ url_for(posts.detail, slug=post.slug) }}">{{ post.title }}</a></h2> <p>{{ post.body|truncate(100) }}</p> <p> {{ post.created_at.strftime(%H:%M %Y-%m-%d) }} | {% with total=post.comments|length %} {{ total }} comment {%- if total > 1 %}s{%- endif -%} {% endwith %} </p> {% endfor %} {% endblock %}
Finally, add templates/posts/detail.html template for the individual posts:

{% extends "base.html" %} {% block page_header %} <div class="page-header"> <h1>{{ post.title }}</h1> </div> {% endblock %}
{% block content %} <p>{{ post.body }}<p> <p>{{ post.created_at.strftime(%H:%M %Y-%m-%d) }}</p> <hr> <h2>Comments</h2> {% if post.comments %} {% for comment in post.comments %} <p>{{ comment.body }}</p> <p><strong>{{ comment.author }}</strong> <small>on {{ comment.created_at.strftime(%H:%M %Y-%m {{ comment.text }} {% endfor %} {% endif %} {% endblock %}
At this point, you can run the python manage.py runserver command again to see your new tumblelog! Go to http://localhost:5000 to see something that resembles the following:
53.2.4 Add Comments to the Blog

In the next step you will provide the facility for readers of the tumblelog to comment on posts. To provide commenting, you will create a form using WTForms that will update the view to handle the form data and update the template to include the form. 53.2. Write a Tumblelog Application with Flask and MongoEngine 695
Handle Comments in the View Begin by updating and refactoring the views.py le so that it can handle the form. Begin by adding the import statement and the DetailView class to this le:
from flask.ext.mongoengine.wtf import model_form ... class DetailView(MethodView): form = model_form(Comment, exclude=[created_at]) def get_context(self, slug): post = Post.objects.get_or_404(slug=slug) form = self.form(request.form) context = { "post": post, "form": form } return context def get(self, slug): context = self.get_context(slug) return render_template(posts/detail.html, **context) def post(self, slug): context = self.get_context(slug) form = context.get(form) if form.validate(): comment = Comment() form.populate_obj(comment) post = context.get(post) post.comments.append(comment) post.save() return redirect(url_for(posts.detail, slug=slug)) return render_template(posts/detail.html, **context)
Note: DetailView extends the default Flask MethodView. This code remains DRY by dening a get_context method to get the default context for both GET and POST requests. On POST, post() validates the comment: if valid, post() appends the comment to the post.
Add Comments to the Templates Finally, you can add the form to the templates, so that readers can create comments. Create a macro for the forms in templates/_forms.html will allow you to reuse the form code:
{% macro render(form) -%} <fieldset> {% for field in form %} {% if field.type in [CSRFTokenField, HiddenField] %}
696
{{ field() }} {% else %} <div class="clearfix {% if field.errors %}error{% endif %}"> {{ field.label }} <div class="input"> {% if field.name == "body" %} {{ field(rows=10, cols=40) }} {% else %} {{ field() }} {% endif %} {% if field.errors or field.help_text %} <span class="help-inline"> {% if field.errors %} {{ field.errors|join( ) }} {% else %} {{ field.help_text }} {% endif %} </span> {% endif %} </div> </div> {% endif %} {% endfor %} </fieldset> {% endmacro %}
Add the comments form to templates/posts/detail.html. Insert an import statement at the top of the page and then output the form after displaying comments:
{% import "_forms.html" as forms %} ... <hr> <h2>Add a comment</h2> <form action="." method="post"> {{ forms.render(form) }} <div class="actions"> <input type="submit" class="btn primary" value="comment"> </div> </form>
Your tumblelogs readers can now comment on your posts! Run python manage.py runserver to see the changes.
697
53.2.5 Add a Site Administration Interface

While you may always add posts using the shell interface as above, in this step you will add an administrative interface for the tumblelog site. To add the administrative interface you will add authentication and an additional view. This tutorial only addresses adding and editing posts: a delete view and detection of slug collisions are beyond the scope of this tutorial. 698 Chapter 53. Python Application Development
Add Basic Authentication For the purposes of this tutorial all we need is a very basic form of authentication. The following example borrows from an example Flask Auth snippet. Create the le auth.py with the following content:
from functools import wraps from flask import request, Response
def check_auth(username, password): """This function is called to check if a username / password combination is valid. """ return username == admin and password == secret
def authenticate(): """Sends a 401 response that enables basic auth""" return Response( Could not verify your access level for that URL.\n You have to login with proper credentials, 401, {WWW-Authenticate: Basic realm="Login Required"})
def requires_auth(f): @wraps(f) def decorated(*args, **kwargs): auth = request.authorization if not auth or not check_auth(auth.username, auth.password): return authenticate() return f(*args, **kwargs) return decorated
Note: This creates a requires_auth decorator: provides basic authentication. Decorate any view that needs authentication with this decorator. The username is admin and password is secret.
Write an Administrative View Create the views and admin blueprint in admin.py. The following view is deliberately generic, to facilitate customization.
from flask import Blueprint, request, redirect, render_template, url_for from flask.views import MethodView from flask.ext.mongoengine.wtf import model_form from tumblelog.auth import requires_auth from tumblelog.models import Post, Comment admin = Blueprint(admin, __name__, template_folder=templates)
class List(MethodView): decorators = [requires_auth] cls = Post
699
def get(self): posts = self.cls.objects.all() return render_template(admin/list.html, posts=posts)
class Detail(MethodView): decorators = [requires_auth] def get_context(self, slug=None): form_cls = model_form(Post, exclude=(created_at, comments)) if slug: post = Post.objects.get_or_404(slug=slug) if request.method == POST: form = form_cls(request.form, inital=post._data) else: form = form_cls(obj=post) else: post = Post() form = form_cls(request.form) context = { "post": post, "form": form, "create": slug is None } return context def get(self, slug): context = self.get_context(slug) return render_template(admin/detail.html, **context) def post(self, slug): context = self.get_context(slug) form = context.get(form) if form.validate(): post = context.get(post) form.populate_obj(post) post.save() return redirect(url_for(admin.index)) return render_template(admin/detail.html, **context)
# Register the urls admin.add_url_rule(/admin/, view_func=List.as_view(index)) admin.add_url_rule(/admin/create/, defaults={slug: None}, view_func=Detail.as_view(create)) admin.add_url_rule(/admin/<slug>/, view_func=Detail.as_view(edit))
Note: Here, the List and Detail views are similar to the frontend of the site; however, requires_auth decorates both views. The Detail view is slightly more complex: to set the context, this view checks for a slug and if there is no slug, Detail uses the view for creating a new post. If a slug exists, Detail uses the view for editing an existing post. In the __init__.py le update the register_blueprints() method to import the new admin blueprint. 700 Chapter 53. Python Application Development
def register_blueprints(app): # Prevents circular imports from tumblelog.views import posts from tumblelog.admin import admin app.register_blueprint(posts) app.register_blueprint(admin)
Create Administrative Templates Similar to the user-facing portion of the site, the administrative section of the application requires three templates: a base template a list view, and a detail view. Create an admin directory for the templates. templates/admin/base.html le:
{% extends "base.html" %} {%- block topbar -%} <div class="topbar" data-dropdown="dropdown"> <div class="fill"> <div class="container"> <h2> <a href="{{ url_for(admin.index) }}" class="brand">My Tumblelog Admin</a> </h2> <ul class="nav secondary-nav"> <li class="menu"> <a href="{{ url_for("admin.create") }}" class="btn primary">Create new post</a> </li> </ul> </div> </div> </div> {%- endblock -%}
Add a simple main index page for the admin in the
List all the posts in the templates/admin/list.html le:

{% extends "admin/base.html" %} {% block content %} <table class="condensed-table zebra-striped"> <thead> <th>Title</th> <th>Created</th> <th>Actions</th> </thead> <tbody> {% for post in posts %} <tr> <th><a href="{{ url_for(admin.edit, slug=post.slug) }}">{{ post.title }}</a></th> <td>{{ post.created_at.strftime(%Y-%m-%d) }}</td> <td><a href="{{ url_for("admin.edit", slug=post.slug) }}" class="btn primary">Edit</a></td> </tr> {% endfor %} </tbody> </table> {% endblock %}
Add a template to create and edit posts in the templates/admin/detail.html le: 53.2. Write a Tumblelog Application with Flask and MongoEngine 701
{% extends "admin/base.html" %} {% import "_forms.html" as forms %} {% block content %} <h2> {% if create %} Add new Post {% else %} Edit Post {% endif %} </h2> <form action="?{{ request.query_string }}" method="post"> {{ forms.render(form) }} <div class="actions"> <input type="submit" class="btn primary" value="save"> <a href="{{ url_for("admin.index") }}" class="btn secondary">Cancel</a> </div> </form> {% endblock %}
The administrative interface is ready for use. Restart the test server (i.e. runserver) so that you can log in to the administrative interface located at http://localhost:5000/admin/. (The username is admin and the password is secret.)
53.2.6 Converting the Blog to a Tumblelog

Currently, the application only supports posts. In this section you will add special post types including: Video, Image and Quote to provide a more traditional tumblelog application. Adding this data requires no migration because MongoEngine supports document inheritance. Begin by refactoring the Post class to operate as a base class and create new classes for the new post types. Update the models.py le to include the code to replace the old Post class:
class Post(db.DynamicDocument): created_at = db.DateTimeField(default=datetime.datetime.now, required=True) title = db.StringField(max_length=255, required=True) slug = db.StringField(max_length=255, required=True) comments = db.ListField(db.EmbeddedDocumentField(Comment))
702
def get_absolute_url(self): return url_for(post, kwargs={"slug": self.slug}) def __unicode__(self): return self.title @property def post_type(self): return self.__class__.__name__ meta = { allow_inheritance: True, indexes: [-created_at, slug], ordering: [-created_at] }
class BlogPost(Post): body = db.StringField(required=True)
class Video(Post): embed_code = db.StringField(required=True)
class Image(Post): image_url = db.StringField(required=True, max_length=255)
class Quote(Post): body = db.StringField(required=True) author = db.StringField(verbose_name="Author Name", required=True, max_length=255)
Note: In the Post class the post_type helper returns the class name, which will make it possible to render the various different post types in the templates. As MongoEngine handles returning the correct classes when fetching Post objects you do not need to modify the interface view logic: only modify the templates. Update the templates/posts/list.html le and change the post output format as follows:
{% if post.body %} {% if post.post_type == Quote %} <blockquote>{{ post.body|truncate(100) }}</blockquote> <p>{{ post.author }}</p> {% else %} <p>{{ post.body|truncate(100) }}</p> {% endif %} {% endif %} {% if post.embed_code %} {{ post.embed_code|safe() }} {% endif %} {% if post.image_url %} <p><img src="{{ post.image_url }}" /><p> {% endif %}
In the templates/posts/detail.html change the output for full posts as follows:
703
{% if post.body %} {% if post.post_type == Quote %} <blockquote>{{ post.body }}</blockquote> <p>{{ post.author }}</p> {% else %} <p>{{ post.body }}</p> {% endif %} {% endif %} {% if post.embed_code %} {{ post.embed_code|safe() }} {% endif %} {% if post.image_url %} <p><img src="{{ post.image_url }}" /><p> {% endif %}
Updating the Administration In this section you will update the administrative interface to support the new post types. Begin by, updating the admin.py le to import the new document models and then update get_context() in the Detail class to dynamically create the correct model form to use:
from tumblelog.models import Post, BlogPost, Video, Image, Quote, Comment # ... class Detail(MethodView): decorators = [requires_auth] # Map post types to models class_map = { post: BlogPost, video: Video, image: Image, quote: Quote, } def get_context(self, slug=None): if slug: post = Post.objects.get_or_404(slug=slug) # Handle old posts types as well cls = post.__class__ if post.__class__ != Post else BlogPost form_cls = model_form(cls, exclude=(created_at, comments)) if request.method == POST: form = form_cls(request.form, inital=post._data) else: form = form_cls(obj=post) else: # Determine which post type we need cls = self.class_map.get(request.args.get(type, post)) post = cls() form_cls = model_form(cls, exclude=(created_at, comments)) form = form_cls(request.form) context = { "post": post, "form": form,
704
"create": slug is None } return context # ...
Update the template/admin/base.html le to create a new post drop down menu in the toolbar:
{% extends "base.html" %} {%- block topbar -%} <div class="topbar" data-dropdown="dropdown"> <div class="fill"> <div class="container"> <h2> <a href="{{ url_for(admin.index) }}" class="brand">My Tumblelog Admin</a> </h2> <ul class="nav secondary-nav"> <li class="menu"> <a href="#" class="menu">Create new</a> <ul class="menu-dropdown"> {% for type in (post, video, image, quote) %} <li><a href="{{ url_for("admin.create", type=type) }}">{{ type|title }}</a></li> {% endfor %} </ul> </li> </ul> </div> </div> </div> {%- endblock -%} {% block js_footer %} <script src="http://ajax.googleapis.com/ajax/libs/jquery/1.7.1/jquery.min.js"></script> <script src="http://twitter.github.com/bootstrap/1.4.0/bootstrap-dropdown.js"></script> {% endblock %}
Now you have a fully edged tumbleblog using Flask and MongoEngine!
705

The complete source code is available on Github: <https://github.com/rozza/ask-tumblelog>
706
Part XIII
Frequently Asked Questions
707
CHAPTER 54
FAQ: MongoDB Fundamentals
Frequently Asked Questions: What kind of database is MongoDB? (page 709) Do MongoDB databases have tables? (page 710) Do MongoDB databases have schemas? (page 710) What languages can I use to work with MongoDB? (page 710) Does MongoDB support SQL? (page 710) What are typical uses for MongoDB? (page 710) Does MongoDB support transactions? (page 711) Does MongoDB require a lot of RAM? (page 711) How do I congure the cache size? (page 711) Does MongoDB require a separate caching layer for application-level caching? (page 711) Does MongoDB handle caching? (page 712) Are writes written to disk immediately, or lazily? (page 712) What language is MongoDB written in? (page 712) What are the limitations of 32-bit versions of MongoDB? (page 712)
This document addresses basic high level questions about MongoDB and its use. If you dont nd the answer youre looking for, check the complete list of FAQs (page 709) or post your question to the MongoDB User Mailing List.
54.1 What kind of database is MongoDB?

MongoDB is a document-oriented DBMS. Think of MySQL but with JSON -like objects comprising the data model, rather than RDBMS tables. Signicantly, MongoDB supports neither joins nor transactions. However, it features secondary indexes, an expressive query language, atomic writes on a per-document level, and fully-consistent reads. Operationally, MongoDB features master-slave replication with automated failover and built-in horizontal scaling via automated range-based partitioning. Note: MongoDB uses BSON , a binary object format similar to, but more expressive than JSON .
709
54.2 Do MongoDB databases have tables?

Instead of tables, a MongoDB database stores its data in collections, which are the rough equivalent of RDBMS tables. A collection holds one or more documents, which corresponds to a record or a row in a relational database table, and each document has one or more elds, which corresponds to a column in a relational database table. Collections have important differences from RDBMS tables. Documents in a single collection may have a unique combination and set of elds. Documents need not have identical elds. You can add a eld to some documents in a collection without adding that eld to all documents in the collection. See also: SQL to MongoDB Mapping Chart (page 1016)
54.3 Do MongoDB databases have schemas?

MongoDB uses dynamic schemas. You can create collections without dening the structure, i.e. the elds or the types of their values, of the documents in the collection. You can change the structure of documents simply by adding new elds or deleting existing ones. Documents in a collection need not have an identical set of elds. In practice, it is common for the documents in a collection to have a largely homogeneous structure; however, this is not a requirement. MongoDBs exible schemas mean that schema migration and augmentation are very easy in practice, and you will rarely, if ever, need to write scripts that perform alter table type operations, which simplies and facilitates iterative software development with MongoDB. See also: SQL to MongoDB Mapping Chart (page 1016)
54.4 What languages can I use to work with MongoDB?

MongoDB client drivers exist for all of the most popular programming languages, and many other ones. See the latest list of drivers for details. See also: MongoDB Drivers and Client Libraries (page 555).
54.5 Does MongoDB support SQL?

No. However, MongoDB does support a rich, ad-hoc query language of its own. See also: Query, Update and Projection Operators (page 763)
54.6 What are typical uses for MongoDB?

MongoDB has a general-purpose design, making it appropriate for a large number of use cases. Examples include content management systems, mobile app, gaming, e-commerce, analytics, archiving, and logging.
710
Chapter 54. FAQ: MongoDB Fundamentals
Do not use MongoDB for systems that require SQL, joins, and multi-object transactions.
54.7 Does MongoDB support transactions?

MongoDB does not provide ACID transactions. However, MongoDB does provide some basic transactional capabilities. Atomic operations are possible within the scope of a single document: that is, we can debit a and credit b as a transaction if they are elds within the same document. Because documents can be rich, some documents contain thousands of elds, with support for testing elds in sub-documents. Additionally, you can make writes in MongoDB durable (the D in ACID). To get durable writes, you must enable journaling, which is on by default in 64-bit builds. You must also issue writes with a write concern of {j: true} to ensure that the writes block until the journal has synced to disk. Users have built successful e-commerce systems using MongoDB, but applications requiring multi-object commits with rollback generally arent feasible.
54.8 Does MongoDB require a lot of RAM?

Not necessarily. Its certainly possible to run MongoDB on a machine with a small amount of free RAM. MongoDB automatically uses all free memory on the machine as its cache. System resource monitors show that MongoDB uses a lot of memory, but its usage is dynamic. If another process suddenly needs half the servers RAM, MongoDB will yield cached memory to the other process. Technically, the operating systems virtual memory subsystem manages MongoDBs memory. This means that MongoDB will use as much free memory as it can, swapping to disk as needed. Deployments with enough memory to t the applications working data set in RAM will achieve the best performance. See also: FAQ: MongoDB Diagnostics (page 755) for answers to additional questions about MongoDB and Memory use.
54.9 How do I congure the cache size?

MongoDB has no congurable cache. MongoDB uses all free memory on the system automatically by way of memorymapped les. Operating systems use the same approach with their le system caches.
54.10 Does MongoDB require application-level caching?
separate
caching
layer
for
No. In MongoDB, a documents representation in the database is similar to its representation in application memory. This means the database already stores the usable form of data, making the data usable in both the persistent store and in the application cache. This eliminates the need for a separate caching layer in the application. This differs from relational databases, where caching data is more expensive. Relational databases must transform data into object representations that applications can read and must store the transformed data in a separate cache: if these transformation from data to application objects require joins, this process increases the overhead related to using the database which increases the importance of the caching layer.
54.7. Does MongoDB support transactions?
711
54.11 Does MongoDB handle caching?

Yes. MongoDB keeps all of the most recently used data in RAM. If you have created indexes for your queries and your working data set ts in RAM, MongoDB serves all queries from memory. MongoDB does not implement a query cache: MongoDB serves all queries directly from the indexes and/or data les.
54.12 Are writes written to disk immediately, or lazily?

Writes are physically written to the journal (page 71) within 100 milliseconds, by default. At that point, the write is durable in the sense that after a pull-plug-from-wall event, the data will still be recoverable after a hard restart. See journalCommitInterval (page 1083) for more information on the journal commit window. While the journal commit is nearly instant, MongoDB writes to the data les lazily. MongoDB may wait to write data to the data les for as much as one minute by default. This does not affect durability, as the journal has enough information to ensure crash recovery. To change the interval for writing to the data les, see syncdelay (page 1086).
54.13 What language is MongoDB written in?

MongoDB is implemented in C++. Drivers and client libraries are typically written in their respective languages, although some drivers use C extensions for better performance.
54.14 What are the limitations of 32-bit versions of MongoDB?

MongoDB uses memory-mapped les (page 745). When running a 32-bit build of MongoDB, the total storage size for the server, including data and indexes, is 2 gigabytes. For this reason, do not deploy MongoDB to production on 32-bit machines. If youre running a 64-bit build of MongoDB, theres virtually no limit to storage size. For production deployments, 64-bit builds and operating systems are strongly recommended. See also: Blog Post: 32-bit Limitations Note: 32-bit builds disable journaling by default because journaling further limits the maximum amount of data that the database can store.
712
Chapter 54. FAQ: MongoDB Fundamentals
CHAPTER 55
FAQ: MongoDB for Application Developers
Frequently Asked Questions: What is a namespace in MongoDB? (page 713) How do you copy all objects from one collection to another? (page 714) If you remove a document, does MongoDB remove it from disk? (page 714) When does MongoDB write updates to disk? (page 714) How do I do transactions and locking in MongoDB? (page 714) How do you aggregate data with MongoDB? (page 715) Why does MongoDB log so many Connection Accepted events? (page 715) Does MongoDB run on Amazon EBS? (page 715) Why are MongoDBs data les so large? (page 715) How do I optimize storage use for small documents? (page 716) When should I use GridFS? (page 716) How does MongoDB address SQL or Query injection? (page 717) BSON (page 717) JavaScript (page 717) Dollar Sign Operator Escaping (page 718) Driver-Specic Issues (page 718) How does MongoDB provide concurrency? (page 718) What is the compare order for BSON types? (page 719) How do I query for elds that have null values? (page 720) Are there any restrictions on the names of Collections? (page 720) How do I isolate cursors from intervening write operations? (page 721) When should I embed documents within other documents? (page 721) Can I manually pad documents to prevent moves during updates? (page 722)
This document answers common questions about application development using MongoDB. If you dont nd the answer youre looking for, check the complete list of FAQs (page 709) or post your question to the MongoDB User Mailing List.
55.1 What is a namespace in MongoDB?

A namespace is the concatenation of the database name and the collection names with a period character in between. Collections are containers for documents that share one or more indexes. Databases are groups of collections stored on disk using a single set of data les.
713
For an example acme.users namespace, acme is the database name and users is the collection name. Period characters can occur in collection names, so that acme.user.history is a valid namespace, with acme as the database name, and user.history as the collection name. While data models like this appear to support nested collections, the collection namespace is at, and there is no difference from the perspective of MongoDB between acme, acme.users, and acme.records.
55.2 How do you copy all objects from one collection to another?
In the mongo (page 1036) shell, you can use the following operation to duplicate the entire collection:
db.people.find().forEach( function(x){db.user.insert(x)} );
Note: Because this process decodes BSON documents to JSON during the copy procedure, documents may incur a loss of type-delity. Consider using mongodump (page 1044) and mongorestore (page 1048) to maintain type delity. Also consider the cloneCollection (page 856) command that may provide some of this functionality.
55.3 If you remove a document, does MongoDB remove it from disk?

Yes. When you use db.collection.remove() (page 944), the object will no longer exist in MongoDBs on-disk data storage.
55.4 When does MongoDB write updates to disk?

MongoDB ushes writes to disk on a regular interval. In the default conguration, MongoDB writes data to the main data les on disk every 60 seconds and commits the journal roughly every 100 milliseconds. These values are congurable with the journalCommitInterval (page 1083) and syncdelay (page 1086). These values represent the maximum amount of time between the completion of a write operation and the point when the write is durable in the journal, if enabled, and when MongoDB ushes data to the disk. In many cases MongoDB and the operating system ush data to disk more frequently, so that the above values resents a theoretical maximum. However, by default, MongoDB uses a lazy strategy to write to disk. This is advantageous in situations where the database receives a thousand increments to an object within one second, MongoDB only needs to ush this data to disk once. In addition to the aforementioned conguration options, you can also use fsync (page 865) and getLastError (page 831) to modify this strategy.
55.5 How do I do transactions and locking in MongoDB?

MongoDB does not have support for traditional locking or complex transactions with rollback. MongoDB aims to be lightweight, fast, and predictable in its performance. This is similar to the MySQL MyISAM autocommit model. By keeping transaction support extremely simple, MongoDB can provide greater performance especially for partitioned or replicated systems with a number of database server processes.
714
Chapter 55. FAQ: MongoDB for Application Developers
MongoDB does have support for atomic operations within a single document. Given the possibilities provided by nested documents, this feature provides support for a large number of use-cases. See also: The Isolate Sequence of Operations (page 571) page.
55.6 How do you aggregate data with MongoDB?

In version 2.1 and later, you can use the new aggregation framework (page 255), with the aggregate (page 808) command. MongoDB also supports map-reduce with the mapReduce (page 814) command, as well as basic aggregation with the group (page 810), count (page 808), and distinct (page 809). commands. See also: The Aggregation (page 253) page.
55.7 Why does MongoDB log so many Connection Accepted events?

If you see a very large number connection and re-connection messages in your MongoDB log, then clients are frequently connecting and disconnecting to the MongoDB server. This is normal behavior for applications that do not use request pooling, such as CGI. Consider using FastCGI, an Apache Module, or some other kind of persistent application server to decrease the connection overhead. If these connections do not impact your performance you can use the run-time quiet (page 1086) option or the command-line option --quiet (page 1022) to suppress these messages from the log.
55.8 Does MongoDB run on Amazon EBS?

Yes. MongoDB users of all sizes have had a great deal of success using MongoDB on the EC2 platform using EBS disks. See also: Amazon EC2
55.9 Why are MongoDBs data les so large?

MongoDB aggressively preallocates data les to reserve space and avoid le system fragmentation. You can use the smallfiles (page 1085) setting to modify the le preallocation strategy. See also: Why are the les in my data directory larger than the data in my database? (page 747)
55.6. How do you aggregate data with MongoDB?
715
55.10 How do I optimize storage use for small documents?

Each MongoDB document contains a certain amount of overhead. This overhead is normally insignicant but becomes signicant if all documents are just a few bytes, as might be the case if the documents in your collection only have one or two elds. Consider the following suggestions and strategies for optimizing storage utilization for these collections: Use the _id eld explicitly. MongoDB clients automatically add an _id eld to each document and generate a unique 12-byte ObjectId for the _id eld. Furthermore, MongoDB always indexes the _id eld. For smaller documents this may account for a signicant amount of space. To optimize storage use, users can specify a value for the _id eld explicitly when inserting documents into the collection. This strategy allows applications to store a value in the _id eld that would have occupied space in another portion of the document. You can store any value in the _id eld, but because this value serves as a primary key for documents in the collection, it must uniquely identify them. If the elds value is not unique, then it cannot serve as a primary key as there would be collisions in the collection. Use shorter eld names. MongoDB stores all eld names in every document. For most documents, this represents a small fraction of the space used by a document; however, for small documents the eld names may represent a proportionally large amount of space. Consider a collection of documents that resemble the following:
{ last_name : "Smith", best_score: 3.9 }
If you shorten the led named last_name to lname and the eld name best_score to score, as follows, you could save 9 bytes per document.
{ lname : "Smith", score : 3.9 }
Shortening eld names reduces expressiveness and does not provide considerable benet on for larger documents and where document overhead is not signicant concern. Shorter eld names do not reduce the size of indexes, because indexes have a predened structure. In general it is not necessary to use short eld names. Embed documents. In some cases you may want to embed documents in other documents and save on the per-document overhead.
55.11 When should I use GridFS?

For documents in a MongoDB collection, you should always use GridFS for storing les larger than 16 MB. In some situations, storing large les may be more efcient in a MongoDB database than on a system-level lesystem. If your lesystem limits the number of les in a directory, you can use GridFS to store as many les as needed. When you want to keep your les and metadata automatically synced and deployed across a number of systems and facilities. When using geographically distributed replica sets (page 396) MongoDB can distribute les and their metadata automatically to a number of mongod (page 1021) instances and facilities. When you want to access information from portions of large les without having to load whole les into memory, you can use GridFS to recall sections of les without reading the entire le into memory.
716
Do not use GridFS if you need to update the content of the entire le atomically. As an alternative you can store multiple versions of each le and specify the current version of the le in the metadata. You can update the metadata eld that indicates latest status in an atomic update after uploading the new version of the le, and later remove previous versions if needed. Furthermore, if your les are all smaller the 16 MB BSON Document Size (page 1105) limit, consider storing the le manually within a single document. You may use the BinData data type to store the binary data. See your drivers (page 555) documentation for details on using BinData. For more information on GridFS, see GridFS (page 196).
55.12 How does MongoDB address SQL or Query injection?

55.12.1 BSON
As a client program assembles a query in MongoDB, it builds a BSON object, not a string. Thus traditional SQL injection attacks are not a problem. More details and some nuances are covered below. MongoDB represents queries as BSON objects. Typically client libraries (page 555) provide a convenient, injection free, process to build these objects. Consider the following C++ example:
BSONObj my_query = BSON( "name" << a_name ); auto_ptr<DBClientCursor> cursor = c.query("tutorial.persons", my_query);
Here, my_query then will have a value such as { name : "Joe" }. If my_query contained special characters, for example ,, :, and {, the query simply wouldnt match any documents. For example, users cannot hijack a query and convert it to a delete.
55.12.2 JavaScript
Note: You can disable all server-side execution of JavaScript, by passing the --noscripting (page 1026) option on the command line or setting noscripting (page 1084) in a conguration le. All of the following MongoDB operations permit you to run arbitrary JavaScript expressions directly on the server: $where (page 775) db.eval() (page 975) mapReduce (page 814) group (page 810) You must exercise care in these cases to prevent users from submitting malicious JavaScript. Fortunately, you can express most queries in MongoDB without JavaScript and for queries that require JavaScript, you can mix JavaScript and non-JavaScript in a single query. Place all the user-supplied elds directly in a BSON eld and pass JavaScript code to the $where (page 775) eld. If you need to pass user-supplied values in a $where (page 775) clause, you may escape these values with the CodeWScope mechanism. When you set user-submitted values as variables in the scope document, you can avoid evaluating them on the database server. If you need to use db.eval() (page 975) with user supplied values, you can either use a CodeWScope or you can supply extra arguments to your function. For instance:
55.12. How does MongoDB address SQL or Query injection?
717
db.eval(function(userVal){...}, user_value);
This will ensure that your application sends user_value to the database server as data rather than code.
55.12.3 Dollar Sign Operator Escaping

Field names in MongoDBs query language have semantic meaning. The dollar sign (i.e $) is a reserved character used to represent operators (page 763) (i.e. $inc (page 784).) Thus, you should ensure that your applications users cannot inject operators into their inputs. In some cases, you may wish to build a BSON object with a user-provided key. In these situations, keys will need to substitute the reserved $ and . characters. Any character is sufcient, but consider using the Unicode full width equivalents: U+FF04 (i.e. $) and U+FF0E (i.e. .). Consider the following example:
BSONObj my_object = BSON( a_key << a_name );
The user may have supplied a $ value in the a_key value. At the same time, my_object might be { $where : "things" }. Consider the following cases: Insert. Inserting this into the database does no harm. The insert process does not evaluate the object as a query. Note: MongoDB client drivers, if properly implemented, check for reserved characters in keys on inserts. Update. The db.collection.update() (page 948) operation permits $ operators in the update argument but does not support the $where (page 775) operator. Still, some users may be able to inject operators that can manipulate a single document only. Therefore your application should escape keys, as mentioned above, if reserved characters are possible. Query Generally this is not a problem for queries that resemble { x : user_obj }: dollar signs are not top level and have no effect. Theoretically it may be possible for the user to build a query themselves. But checking the user-submitted content for $ characters in key names may help protect against this kind of injection.
55.12.4 Driver-Specic Issues

See the PHP MongoDB Driver Security Notes page in the PHP driver documentation for more information
55.13 How does MongoDB provide concurrency?

MongoDB implements a readers-writer lock. This means that at any one time, only one client may be writing or any number of clients may be reading, but that reading and writing cannot occur simultaneously. In standalone and replica sets the locks scope applies to a single mongod (page 1021) instance or primary instance. In a sharded cluster, locks apply to each individual shard, not to the whole cluster. For more information, see FAQ: Concurrency (page 727).
718
55.14 What is the compare order for BSON types?

MongoDB permits documents within a single collection to have elds with different BSON types. For instance, the following documents may exist within a single collection.
{ x: "string" } { x: 42 }
When comparing values of different BSON types, MongoDB uses the following comparison order, from lowest to highest: 1. MinKey (internal type) 2. Null 3. Numbers (ints, longs, doubles) 4. Symbol, String 5. Object 6. Array 7. BinData 8. ObjectID 9. Boolean 10. Date, Timestamp 11. Regular Expression 12. MaxKey (internal type) Note: MongoDB treats some types as equivalent for comparison purposes. For instance, numeric types undergo conversion before comparison. Consider the following mongo (page 1036) example:
db.test.insert( db.test.insert( db.test.insert( db.test.insert( {x {x {x {x : : : : 3 } ); 2.9 } ); new Date() } ); true } );
db.test.find().sort({x:1}); { "_id" : ObjectId("4b03155dce8de6586fb002c7"), { "_id" : ObjectId("4b03154cce8de6586fb002c6"), { "_id" : ObjectId("4b031566ce8de6586fb002c9"), { "_id" : ObjectId("4b031563ce8de6586fb002c8"),
"x" "x" "x" "x"
: : : :
2.9 } 3 } true } "Tue Nov 17 2009 16:28:03 GMT-0500 (EST)" }
The $type (page 773) operator provides access to BSON type comparison in the MongoDB query syntax. See the documentation on BSON types and the $type (page 773) operator for additional information. Warning: Storing values of the different types in the same eld in a collection is strongly discouraged. See also: The Tailable Cursors (page 568) page for an example of a C++ use of MinKey.
55.14. What is the compare order for BSON types?
719
55.15 How do I query for elds that have null values?

Fields in a document may store null values, as in a notional collection, test, with the following documents:
{ _id: 1, cancelDate: null } { _id: 2 }
Different query operators treat null values differently: The { cancelDate : null } query matches documents that either contains the cancelDate eld whose value is null or that do not contain the cancelDate eld:
db.test.find( { cancelDate: null } )
The query returns both documents:

{ "_id" : 1, "cancelDate" : null } { "_id" : 2 }
The { cancelDate : { $type: 10 } } query matches documents that contains the cancelDate eld whose value is null only; i.e. the value of the cancelDate eld is of BSON Type Null (i.e. 10) :
db.test.find( { cancelDate : { $type: 10 } } )
The query returns only the document that contains the null value:
{ "_id" : 1, "cancelDate" : null }
The { cancelDate : cancelDate eld:
{ $exists:
false } } query matches documents that do not contain the
db.test.find( { cancelDate : { $exists: false } } )
The query returns only the document that does not contain the cancelDate eld:
{ "_id" : 2 }
See also: The reference documentation for the $type (page 773) and $exists (page 772) operators.
55.16 Are there any restrictions on the names of Collections?

Collection names can be any UTF-8 string with the following exceptions: A collection name should begin with a letter or an underscore. The empty string ("") is not a valid collection name. Collection names cannot contain the $ character. (version 2.2 only) Collection names cannot contain the null character: \0 Do not name a collection using the system. prex. MongoDB reserves system. for system collections, such as the system.indexes collection. The maximum size of a collection name is 128 characters, including the name of the database. However, for maximum exibility, collections should have names less than 80 characters.
720
If your collection name includes special characters, such as the underscore character, then to access the collection use the db.getCollection() (page 978) method or a similar method for your driver. Example To create a collection _foo and insert the { a : 1 } document, use the following operation:
db.getCollection("_foo").insert( { a : 1 } )
To perform a query, use the find() (page 924) method, in as the following:
db.getCollection("_foo").find()
55.17 How do I isolate cursors from intervening write operations?

MongoDB cursors can return the same document more than once in some situations. (page 964) method on a cursor to isolate the operation for a very specic case.
1
You can use the snapshot()
snapshot() (page 964) traverses the index on the _id eld and guarantees that the query will return each document (with respect to the value of the _id eld) no more than once. 2 The snapshot() (page 964) does not guarantee that the data returned by the query will reect a single moment in time nor does it provide isolation from insert or delete operations. Warning: You cannot use snapshot() (page 964) with sharded collections. You cannot use snapshot() (page 964) with sort() (page 965) or hint() (page 958) cursor methods. As an alternative, if your collection has a eld or elds that are never modied, you can use a unique index on this eld or these elds to achieve a similar result as the snapshot() (page 964). Query with hint() (page 958) to explicitly force the query to use that index.
55.18 When should I embed documents within other documents?

When modeling data in MongoDB (page 233), embedding is frequently the choice for: contains relationships between entities. one-to-many relationships when the many objects always appear with or are viewed in the context of their parents. You should also consider embedding for performance reasons if you have a collection with a large number of small documents. Nevertheless, if small, separate documents represent the natural model for the data, then you should maintain that model. If, however, you can group these small documents by some logical relationship and you frequently retrieve the documents by this grouping, you might consider rolling-up the small documents into larger documents that contain an array of subdocuments. Keep in mind that if you often only need to retrieve a subset of the documents within the group, then rolling-up the documents may not provide better performance.
1 As a cursor returns documents other operations may interleave with the query: if some of these operations are updates (page 219) that cause the document to move (in the case of a table scan, caused by document growth,) or that change the indexed eld on the index used by the query; then the cursor will return the same document more than once. 2 MongoDB does not permit changes to the value of the _id eld; it is not possible for a cursor that transverses this index to pass the same document more than once.
55.17. How do I isolate cursors from intervening write operations?
721
Rolling up these small documents into logical groupings means that queries to retrieve a group of documents involve sequential reads and fewer random disk accesses. Additionally, rolling up documents and moving common elds to the larger document benet the index on these elds. There would be fewer copies of the common elds and there would be fewer associated key entries in the corresponding index. See Indexing Overview (page 329) for more information on indexes.
55.19 Can I manually pad documents to prevent moves during updates?

An update can cause a document to move on disk if the document grows in size. To minimize document movements, MongoDB uses padding (page 181). You should not have to pad manually because MongoDB adds padding automatically (page 181) and can adaptively adjust the amount of padding added to documents to prevent document relocations following updates. You can change the default paddingFactor (page 874) calculation by using the collMod (page 857) command with the usePowerOf2Sizes (page 857) ag. The usePowerOf2Sizes (page 857) ag ensures that MongoDB allocates document space in sizes that are powers of 2, which helps ensure that MongoDB can efciently reuse free pace created by document deletion or relocation. However, in those exceptions where you must pad manually, you can use the strategy of rst adding a temporary eld to a document and then $unset (page 788) the eld, as in the following example:
var myTempPadding = [ "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"]; db.myCollection.insert( { _id: 5, paddingField: myTempPadding } ); db.myCollection.update( { _id: 5 }, { $unset: { paddingField: "" } } ) db.myCollection.update( { _id: 5 }, { $set: { realField: "Some text that I might have needed padding for" } } )
See also: Padding Factor (page 181)
722
CHAPTER 56
FAQ: The mongo Shell
Frequently Asked Questions: How can I enter multi-line operations in the mongo shell? (page 723) How can I access different databases temporarily? (page 723) Does the mongo shell support tab completion and other keyboard shortcuts? (page 724) How can I customize the mongo shell prompt? (page 724) Can I edit long shell operations with an external text editor? (page 724)
56.1 How can I enter multi-line operations in the mongo shell?

If you end a line with an open parenthesis ((), an open brace ({), or an open bracket ([), then the subsequent lines start with ellipsis ("...") until you enter the corresponding closing parenthesis ()), the closing brace (}) or the closing bracket (]). The mongo (page 1036) shell waits for the closing parenthesis, closing brace, or the closing bracket before evaluating the code, as in the following example:
> if ( x > 0 ) { ... count++; ... print (x); ... }
You can exit the line continuation mode if you enter two blank lines, as in the following example:
> if (x > 0 ... ... >
56.2 How can I access different databases temporarily?

You can use db.getSiblingDB() (page 980) method to access another database without switching databases, as in the following example which rst switches to the test database and then accesses the sampleDB database from the test database:
use test db.getSiblingDB(sampleDB).getCollectionNames();
723
56.3 Does the mongo shell support tab completion and other keyboard shortcuts?
The mongo (page 1036) shell supports keyboard shortcuts. For example, Use the up/down arrow keys to scroll through command history. See .dbshell (page 1038) documentation for more information on the .dbshell le. Use <Tab> to autocomplete or to list the completion possibilities, as in the following example which uses <Tab> to complete the method name starting with the letter c:
db.myCollection.c<Tab>
Because there are many collection methods starting with the letter c, the <Tab> will list the various methods that start with c. For a full list of the shortcuts, see Shell Keyboard Shortcuts (page 1039)
56.4 How can I customize the mongo shell prompt?

New in version 1.9. You can change the mongo (page 1036) shell prompt by setting the prompt variable. This makes it possible to display additional information in the prompt. Set prompt to any string or arbitrary JavaScript code that returns a string, consider the following examples: Set the shell prompt to display the hostname and the database issued:
var host = db.serverStatus().host; var prompt = function() { return db+"@"+host+"> "; }
The mongo (page 1036) shell prompt should now reect the new prompt:
test@my-machine.local>
Set the shell prompt to display the database statistics:
var prompt = function() { return "Uptime:"+db.serverStatus().uptime+" Documents:"+db.stats().objects+" > " }
The mongo (page 1036) shell prompt should now reect the new prompt:
Uptime:1052 Documents:25024787 >
You can add the logic for the prompt in the .mongorc.js (page 1039) le to set the prompt each time you start up the mongo (page 1036) shell.
56.5 Can I edit long shell operations with an external text editor?
You can use your own editor in the mongo (page 1036) shell by setting the EDITOR (page 1039) environment variable before starting the mongo (page 1036) shell. Once in the mongo (page 1036) shell, you can edit with the specied editor by typing edit <variable> or edit <function>, as in the following example: 1. Set the EDITOR (page 1039) variable from the command line prompt:
724
Chapter 56. FAQ: The mongo Shell
EDITOR=vim
2. Start the mongo (page 1036) shell:

mongo
3. Dene a function myFunction:

function myFunction () { }
4. Edit the function using your editor:

edit myFunction
The command should open the vim edit session. Remember to save your changes. 5. Type myFunction to see the function denition:
myFunction
The result should be the changes from your saved edit:

function myFunction() { print("This was edited"); }
56.5. Can I edit long shell operations with an external text editor?
725
726
Chapter 56. FAQ: The mongo Shell
CHAPTER 57
FAQ: Concurrency
Frequently Asked Questions: What type of locking does MongoDB use? (page 727) How granular are locks in MongoDB? (page 728) How do I see the status of locks on my mongod (page 1021) instances? (page 728) Does a read or write operation ever yield the lock? (page 728) Which operations lock the database? (page 728) Which administrative commands lock the database? (page 729) Does a MongoDB operation ever lock more than one database? (page 730) How does sharding affect concurrency? (page 730) How does concurrency affect a replica set primary? (page 730) How does concurrency affect secondaries? (page 730) What kind of concurrency does MongoDB provide for JavaScript operations? (page 730)
Changed in version 2.2. MongoDB allows multiple clients to read and write a single corpus of data using a locking system to ensure that all clients receive a consistent view of the data and to prevent multiple applications from modifying the exact same pieces of data at the same time. Locks help guarantee that all writes to a single document occur either in full or not at all. See also: Presentation on Concurrency and Internals in 2.2
57.1 What type of locking does MongoDB use?

MongoDB uses a readers-writer 1 lock that allows concurrent reads access to a database but gives exclusive access to a single write operation. When a read lock exists, many read operations may use this lock. However, when a write lock exists, a single write operation holds the lock exclusively, and no other read or write operations may share the lock. Locks are writer greedy, which means writes have preference over reads. When both a read and write are waiting for a lock, MongoDB grants the lock to the write.
1 You may be familiar with a readers-writer lock as multi-reader or shared exclusive lock. See the Wikipedia page on Readers-Writer Locks for more information.
727
57.2 How granular are locks in MongoDB?

Changed in version 2.2. Beginning with version 2.2, MongoDB implements locks on a per-database basis for most read and write operations. Some global operations, typically short lived operations involving multiple databases, still require a global instance wide lock. Before 2.2, there is only one global lock per mongod (page 1021) instance. For example, if you have six databases and one takes a write lock, the other ve are still available for read and write.
57.3 How do I see the status of locks on my mongod instances?

For reporting on lock utilization information on locks, use any of the following methods: db.serverStatus() (page 984), db.currentOp() (page 971), mongotop (page 1067), mongostat (page 1063), and/or the MongoDB Monitoring Service (MMS) Specically, the locks (page 890) document in the output of serverStatus (page 889), or the locks (page 973) eld in the current operation reporting (page 971) provides insight into the type of locks and amount of lock contention in your mongod (page 1021) instance. To terminate an operation, use db.killOp() (page 981).
57.4 Does a read or write operation ever yield the lock?

New in version 2.0. Read and write operations will yield their locks if the mongod (page 1021) receives a page fault or fetches data that is unlikely to be in memory. Yielding allows other operations that only need to access documents that are already in memory to complete while mongod (page 1021) loads documents into memory. Additionally, write operations that affect multiple documents (i.e. update() (page 948) with the multi parameter,) will yield periodically to allow read operations during these long write operations. Similarly, long running read locks will yield periodically to ensure that write operations have the opportunity to complete. Changed in version 2.2: The use of yielding expanded greatly in MongoDB 2.2. Including the yield for page fault. MongoDB tracks the contents of memory and predicts whether data is available before performing a read. If MongoDB predicts that the data is not in memory a read operation yields its lock while MongoDB loads the data to memory. Once data is available in memory, the read will reacquire the lock to complete the operation.
57.5 Which operations lock the database?

Changed in version 2.2. The following table lists common database operations and the types of locks they use.
728
Chapter 57. FAQ: Concurrency
Operation Issue a query Get more data from a cursor Insert data Remove data Update data Map-reduce Create an index db.eval() (page 975) eval (page 823) aggregate() (page 918)
Lock Type Read lock Read lock Write lock Write lock Write lock Read lock and write lock, unless operations are specied as non-atomic. Portions of map-reduce jobs can run concurrently. Building an index in the foreground, which is the default, locks the database for extended periods of time. Write lock. db.eval() (page 975) blocks all other JavaScript processes. Write lock. If used with the nolock lock option, the eval (page 823) option does not take a write lock and cannot write data to the database. Read lock
57.6 Which administrative commands lock the database?

Certain administrative commands can exclusively lock the database for extended periods of time. In some deployments, for large databases, you may consider taking the mongod (page 1021) instance ofine so that clients are not affected. For example, if a mongod (page 1021) is part of a replica set, take the mongod (page 1021) ofine and let other members of the set service load while maintenance is in progress. The following administrative operations require an exclusive (i.e. write) lock on the database for extended periods: db.collection.ensureIndex() (page 921), when issued without setting background to true, reIndex (page 867), compact (page 859), db.repairDatabase() (page 984), db.createCollection() (page 970), when creating a very large (i.e. many gigabytes) capped collection, db.collection.validate() (page 950), and db.copyDatabase() (page 970). This operation may lock all databases. See Does a MongoDB operation ever lock more than one database? (page 730). The following administrative commands lock the database but only hold the lock for a very short time: db.collection.dropIndex(), db.getLastError() (page 978), db.isMaster() (page 981), rs.status() (page 990) (i.e. replSetGetStatus (page 840),) db.serverStatus() (page 984), db.auth() (page 968), and db.addUser() (page 967).
57.6. Which administrative commands lock the database?
729
57.7 Does a MongoDB operation ever lock more than one database?
The following MongoDB operations lock multiple databases: db.copyDatabase() (page 970) must lock the entire mongod (page 1021) instance at once. Journaling, which is an internal operation, locks all databases for short intervals. All databases share a single journal. User authentication (page 135) locks the admin database as well as the database the user is accessing. All writes to a replica sets primary lock both the database receiving the writes and the local database. The lock for the local database allows the mongod (page 1021) to write to the primarys oplog.
57.8 How does sharding affect concurrency?

Sharding improves concurrency by distributing collections over multiple mongod (page 1021) instances, allowing shard servers (i.e. mongos (page 1032) processes) to perform any number of operations concurrently to the various downstream mongod (page 1021) instances. Each mongod (page 1021) instance is independent of the others in the shard cluster and uses the MongoDB readerswriter lock (page 727)). The operations on one mongod (page 1021) instance do not block the operations on any others.
57.9 How does concurrency affect a replica set primary?

In replication, when MongoDB writes to a collection on the primary, MongoDB also writes to the primarys oplog, which is a special collection in the local database. Therefore, MongoDB must lock both the collections database and the local database. The mongod (page 1021) must lock both databases at the same time keep both data consistent and ensure that write operations, even with replication, are all-or-nothing operations.
57.10 How does concurrency affect secondaries?

In replication, MongoDB does not apply writes serially to secondaries. Secondaries collect oplog entries in batches and then apply those batches in parallel. Secondaries do not allow reads while applying the write operations, and apply write operations in the order that they appear in the oplog. MongoDB can apply several writes in parallel on replica set secondaries, in two phases: 1. During the rst prefer phase, under a read lock, the mongod (page 1021) ensures that all documents affected by the operations are in memory. During this phase, other clients may execute queries against this member. 2. A thread pool using write locks applies all write operations in the batch as part of a coordinated write phase.
57.11 What kind of concurrency does MongoDB provide for JavaScript operations?
Changed in version 2.4: The V8 JavaScript engine added in 2.4 allows multiple JavaScript operations to run at the same time. Prior to 2.4, a single mongod (page 1021) could only run a single JavaScript operation at once.
730
Chapter 57. FAQ: Concurrency
CHAPTER 58
FAQ: Sharding with MongoDB
Frequently Asked Questions: Is sharding appropriate for a new deployment? (page 732) How does sharding work with replication? (page 732) Can I change the shard key after sharding a collection? (page 732) What happens to unsharded collections in sharded databases? (page 732) How does MongoDB distribute data across shards? (page 732) What happens if a client updates a document in a chunk during a migration? (page 733) What happens to queries if a shard is inaccessible or slow? (page 733) How does MongoDB distribute queries among shards? (page 733) How does MongoDB sort queries in sharded environments? (page 733) How does MongoDB ensure unique _id eld values when using a shard key other than _id? (page 734) Ive enabled sharding and added a second shard, but all the data is still on one server. Why? (page 734) Is it safe to remove old les in the moveChunk directory? (page 734) How does mongos use connections? (page 734) Why does mongos hold connections open? (page 734) Where does MongoDB report on connections used by mongos? (page 735) What does writebacklisten in the log mean? (page 735) How should administrators deal with failed migrations? (page 735) What is the process for moving, renaming, or changing the number of cong servers? (page 735) When do the mongos servers detect cong server changes? (page 735) Is it possible to quickly update mongos servers after updating a replica set conguration? (page 735) What does the maxConns setting on mongos do? (page 736) How do indexes impact queries in sharded systems? (page 736) Can shard keys be randomly generated? (page 736) Can shard keys have a non-uniform distribution of values? (page 736) Can you shard on the _id eld? (page 736) Can shard key be in ascending order, like dates or timestamps? (page 737) What do moveChunk commit failed errors mean? (page 737) How does draining a shard affect the balancing of uneven chunk distribution? (page 737)
This document answers common questions about horizontal scaling using MongoDBs sharding. If you dont nd the answer youre looking for, check the complete list of FAQs (page 709) or post your question to the MongoDB User Mailing List.
731
58.1 Is sharding appropriate for a new deployment?

Sometimes. If your data set ts on a single server, you should begin with an unsharded deployment. Converting an unsharded database to a sharded cluster is easy and seamless, so there is little advantage in conguring sharding while your data set is small. Still, all production deployments should use replica sets to provide high availability and disaster recovery.
58.2 How does sharding work with replication?

To use replication with sharding, deploy each shard as a replica set.
58.3 Can I change the shard key after sharding a collection?

No. There is no automatic support in MongoDB for changing a shard key after sharding a collection. This reality underscores the important of choosing a good shard key (page 485). If you must change a shard key after sharding a collection, the best option is to: dump all data from MongoDB into an external format. drop the original sharded collection. congure sharding using a more ideal shard key. pre-split (page 522) the shard key range to ensure initial even distribution. restore the dumped data into MongoDB. See shardCollection (page 851), sh.shardCollection() (page 1000), the Shard Key (page 493) section in the Sharded Cluster Internals (page 493) document, Deploy a Sharded Cluster (page 503), and SERVER-4000 for more information.
58.4 What happens to unsharded collections in sharded databases?

In the current implementation, all databases in a sharded cluster have a primary shard. All unsharded collection within that database will reside on the same shard.
58.5 How does MongoDB distribute data across shards?

Sharding must be specically enabled on a collection. After enabling sharding on the collection, MongoDB will assign various ranges of collection data to the different shards in the cluster. The cluster automatically corrects imbalances between shards by migrating ranges of data from one shard to another.
732
Chapter 58. FAQ: Sharding with MongoDB
58.6 What happens if a client updates a document in a chunk during a migration?

The mongos (page 1032) routes the operation to the old shard, where it will succeed immediately. Then the shard mongod (page 1021) instances will replicate the modication to the new shard before the sharded cluster updates that chunks ownership, which effectively nalizes the migration process.
58.7 What happens to queries if a shard is inaccessible or slow?

If a shard is inaccessible or unavailable, queries will return with an error. However, a client may set the partial query bit, which will then return results from all available shards, regardless of whether a given shard is unavailable. If a shard is responding slowly, mongos (page 1032) will merely wait for the shard to return results.
58.8 How does MongoDB distribute queries among shards?

Changed in version 2.0. The exact method for distributing queries to shards in a cluster depends on the nature of the query and the conguration of the sharded cluster. Consider a sharded collection, using the shard key user_id, that has last_login and email attributes: For a query that selects one or more values for the user_id key: mongos (page 1032) determines which shard or shards contains the relevant data, based on the cluster metadata, and directs a query to the required shard or shards, and returns those results to the client. For a query that selects user_id and also performs a sort: mongos (page 1032) can make a straightforward translation of this operation into a number of queries against the relevant shards, ordered by user_id. When the sorted queries return from all shards, the mongos (page 1032) merges the sorted results and returns the complete result to the client. For queries that select on last_login: These queries must run on all shards: mongos (page 1032) must parallelize the query over the shards and perform a merge-sort on the email of the documents found.
58.9 How does MongoDB sort queries in sharded environments?

If you call the cursor.sort() (page 965) method on a query in a sharded environment, the mongod (page 1021) for each shard will sort its results, and the mongos (page 1032) merges each shards results before returning them to the client.
58.6. What happens if a client updates a document in a chunk during a migration?
733
58.10 How does MongoDB ensure unique _id eld values when using a shard key other than _id?
If you do not use _id as the shard key, then your application/client layer must be responsible for keeping the _id eld unique. It is problematic for collections to have duplicate _id values. If youre not sharding your collection by the _id eld, then you should be sure to store a globally unique identier in that eld. The default BSON ObjectID (page 194) works well in this case.
58.11 Ive enabled sharding and added a second shard, but all the data is still on one server. Why?
First, ensure that youve declared a shard key for your collection. Until you have congured the shard key, MongoDB will not create chunks, and sharding will not occur. Next, keep in mind that the default chunk size is 64 MB. As a result, in most situations, the collection needs to have at least 64 MB of data before a migration will occur. Additionally, the system which balances chunks among the servers attempts to avoid superuous migrations. Depending on the number of shards, your shard key, and the amount of data, systems often require at least 10 chunks of data to trigger migrations. You can run db.printShardingStatus() (page 983) to see all the chunks present in your cluster.
58.12 Is it safe to remove old les in the moveChunk directory?

Yes. mongod (page 1021) creates these les as backups during normal shard balancing operations. Once these migrations are complete, you may delete these les.
58.13 How does mongos use connections?

Each client maintains a connection to a mongos (page 1032) instance. Each mongos (page 1032) instance maintains a pool of connections to the members of a replica set supporting the sharded cluster. Clients use connections between mongos (page 1032) and mongod (page 1021) instances one at a time. Requests are not multiplexed or pipelined. When client requests complete, the mongos (page 1032) returns the connection to the pool. See the System Resource Utilization (page 109) section of the Linux ulimit Settings (page 109) document.
58.14 Why does mongos hold connections open?

mongos (page 1032) uses a set of connection pools to communicate with each shard. These pools do not shrink when the number of clients decreases. This can lead to an unused mongos (page 1032) with a large number of open connections. If the mongos (page 1032) is no longer in use, it is safe to restart the process to close existing connections.
734
58.15 Where does MongoDB report on connections used by mongos?

Connect to the mongos (page 1032) with the mongo (page 1036) shell, and run the following command:
db._adminCommand("connPoolStats");
58.16 What does writebacklisten in the log mean?

The writeback listener is a process that opens a long poll to relay writes back from a mongod (page 1021) or mongos (page 1032) after migrations to make sure they have not gone to the wrong server. The writeback listener sends writes back to the correct server if necessary. These messages are a key part of the sharding infrastructure and should not cause concern.
58.17 How should administrators deal with failed migrations?

Failed migrations require no administrative intervention. Chunk moves are consistent and deterministic. If a migration fails to complete for some reason, the cluster will retry the operation. When the migration completes successfully, the data will reside only on the new shard.
58.18 What is the process for moving, renaming, or changing the number of cong servers?
See Sharded Cluster Administration (page 503) for information on migrating and replacing cong servers.
58.19 When do the mongos servers detect cong server changes?

mongos (page 1032) instances maintain a cache of the cong database that holds the metadata for the sharded cluster. This metadata includes the mapping of chunks to shards. mongos (page 1032) updates its cache lazily by issuing a request to a shard and discovering that its metadata is out of date. There is no way to control this behavior from the client, but you can run the flushRouterConfig (page 848) command against any mongos (page 1032) to force it to refresh its cache.
58.20 Is it possible to quickly update mongos servers after updating a replica set conguration?
The mongos (page 1032) instances will detect these changes without intervention over time. However, if you want to force the mongos (page 1032) to reload its conguration, run the flushRouterConfig (page 848) command against to each mongos (page 1032) directly.
58.15. Where does MongoDB report on connections used by mongos?
735
58.21 What does the maxConns setting on mongos do?

The maxConns (page 1079) option limits the number of connections accepted by mongos (page 1032). If your client driver or application creates a large number of connections but allows them to time out rather than closing them explicitly, then it might make sense to limit the number of connections at the mongos (page 1032) layer. Set maxConns (page 1079) to a value slightly higher than the maximum number of connections that the client creates, or the maximum size of the connection pool. This setting prevents the mongos (page 1032) from causing connection spikes on the individual shards. Spikes like these may disrupt the operation and memory allocation of the sharded cluster.
58.22 How do indexes impact queries in sharded systems?

If the query does not include the shard key, the mongos (page 1032) must send the query to all shards as a scatter/gather operation. Each shard will, in turn, use either the shard key index or another more efcient index to fulll the query. If the query includes multiple sub-expressions that reference the elds indexed by the shard key and the secondary index, the mongos (page 1032) can route the queries to a specic shard and the shard will use the index that will allow it to fulll most efciently. See this presentation for more information.
58.23 Can shard keys be randomly generated?

Shard keys can be random. Random keys ensure optimal distribution of data across the cluster. Sharded clusters, attempt to route queries to specic shards when queries include the shard key as a parameter, because these directed queries are more efcient. In many cases, random keys can make it difcult to direct queries to specic shards.
58.24 Can shard keys have a non-uniform distribution of values?

Yes. There is no requirement that documents be evenly distributed by the shard key. However, documents that have the shard key must reside in the same chunk and therefore on the same server. If your sharded data set has too many documents with the exact same shard key you will not be able to distribute those documents across your sharded cluster.
58.25 Can you shard on the _id eld?

You can use any eld for the shard key. The _id eld is a common shard key. Be aware that ObjectId() values, which are the default value of the _id eld, increment as a timestamp. As a result, when used as a shard key, all new documents inserted into the collection will initially belong to the same chunk on a single shard. Although the system will eventually divide this chunk and migrate its contents to distribute data more evenly, at any moment the cluster can only direct insert operations at a single shard. This can limit the throughput of inserts. If most of your write operations are updates or read operations rather than inserts, this limitation should not impact your performance. However, if you have a high insert volume, this may be a limitation.
736
58.26 Can shard key be in ascending order, like dates or timestamps?

If you insert documents with monotonically increasing shard keys, all inserts will initially belong to the same chunk on a single shard. Although the system will eventually divide this chunk and migrate its contents to distribute data more evenly, at any moment the cluster can only direct insert operations at a single shard. This can limit the throughput of inserts. If most of your write operations are updates or read operations rather than inserts, this limitation should not impact your performance. However, if you have a high insert volume, a monotonically increasing shard key may be a limitation. To address this issue, you can use a eld with a value that stores the hash of a key with an ascending value. Changed in version 2.4: You can use a hashed index (page 335) and hashed shard key or you can compute and maintain this hashed value in your application.
58.27 What do moveChunk commit failed errors mean?

Consider the following error message:
ERROR: moveChunk commit failed: version is at <n>|<nn> instead of <N>|<NN>" and "ERROR: TERMINATING"
mongod (page 1021) issues this message if, during a chunk migration (page 499), the shard could not connect to the cong database to update chunk information at the end of the migration process. If the shard cannot update the cong database after moveChunk (page 849), the cluster will have an inconsistent view of all chunks. In these situations, the primary member of the shard will terminate itself to prevent data inconsistency. If the secondary member can access the cong database, the shards data will be accessible after an election. Administrators will need to resolve the chunk migration failure independently. If you encounter this issue, contact the MongoDB User Group or 10gen support to address this issue.
58.28 How does draining a shard affect the balancing of uneven chunk distribution?
The sharded cluster balancing process controls both migrating chunks from decommissioned shards (i.e. draining,) and normal cluster balancing activities. Consider the following behaviors for different versions of MongoDB in situations where you remove a shard in a cluster with an uneven chunk distribution: After MongoDB 2.2, the balancer rst removes the chunks from the draining shard and then balances the remaining uneven chunk distribution. Before MongoDB 2.2, the balancer handles the uneven chunk distribution and then removes the chunks from the draining shard.
58.26. Can shard key be in ascending order, like dates or timestamps?
737
738
CHAPTER 59
FAQ: Replica Sets and Replication in MongoDB
Frequently Asked Questions: What kinds of replication does MongoDB support? (page 739) What do the terms primary and master mean? (page 739) What do the terms secondary and slave mean? (page 740) How long does replica set failover take? (page 740) Does replication work over the Internet and WAN connections? (page 740) Can MongoDB replicate over a noisy connection? (page 740) What is the preferred replication method: master/slave or replica sets? (page 741) What is the preferred replication method: replica sets or replica pairs? (page 741) Why use journaling if replication already provides data redundancy? (page 741) Are write operations durable if write concern does not acknowledge writes? (page 741) How many arbiters do replica sets need? (page 741) What information do arbiters exchange with the rest of the replica set? (page 742) Which members of a replica set vote in elections? (page 742) Do hidden members vote in replica set elections? (page 742) Is it normal for replica set members to use different amounts of disk space? (page 743)
This document answers common questions about database replication in MongoDB. If you dont nd the answer youre looking for, check the complete list of FAQs (page 709) or post your question to the MongoDB User Mailing List.
59.1 What kinds of replication does MongoDB support?

MongoDB supports master-slave replication and a variation on master-slave replication known as replica sets. Replica sets are the recommended replication topology.
59.2 What do the terms primary and master mean?

Primary and master nodes are the nodes that can accept writes. MongoDBs replication is single-master: only one node can accept write operations at a time. In a replica set, if the current primary node fails or becomes inaccessible, the other members can autonomously elect one of the other members of the set to be the new primary.
739
By default, clients send all reads to the primary; however, read preference is congurable at the client level on a per-connection basis, which makes it possible to send reads to secondary nodes instead.
59.3 What do the terms secondary and slave mean?

Secondary and slave nodes are read-only nodes that replicate from the primary. Replication operates by way of an oplog, from which secondary/slave members apply new operations to themselves. This replication process is asynchronous, so secondary/slave nodes may not always reect the latest writes to the primary. But usually, the gap between the primary and secondary nodes is just few milliseconds on a local network connection.
59.4 How long does replica set failover take?

It varies, but a replica set will select a new primary within a minute. It may take 10-30 seconds for the members of a replica set to declare a primary inaccessible. This triggers an election. During the election, the cluster is unavailable for writes. The election itself may take another 10-30 seconds. Note: Eventually consistent reads, like the ones that will return from a replica set are only possible with a write concern that permits reads from secondary members.
59.5 Does replication work over the Internet and WAN connections?
Yes. For example, a deployment may maintain a primary and secondary in an East-coast data center along with a secondary member for disaster recovery in a West-coast data center. See also: Deploy a Geographically Distributed Replica Set (page 429)
59.6 Can MongoDB replicate over a noisy connection?

Yes, but not without connection failures and the obvious latency. Members of the set will attempt to reconnect to the other members of the set in response to networking aps. This does not require administrator intervention. However, if the network connections among the nodes in the replica set are very slow, it might not be possible for the members of the node to keep up with the replication. If the TCP connection between the secondaries and the primary instance breaks, a replica set will automatically elect one of the secondary members of the set as primary.
740
Chapter 59. FAQ: Replica Sets and Replication in MongoDB
59.7 What is the preferred replication method: replica sets?

New in version 1.8.
master/slave or
Replica sets are the preferred replication mechanism in MongoDB. However, if your deployment requires more than 12 nodes, you must use master/slave replication.
59.8 What is the preferred replication method: replica sets or replica pairs?
Deprecated since version 1.6. Replica sets replaced replica pairs in version 1.6. Replica sets are the preferred replication mechanism in MongoDB.
59.9 Why use journaling if replication already provides data redundancy?

Journaling facilitates faster crash recovery. Prior to journaling, crashes often required database repairs (page 868) or full data resync. Both were slow, and the rst was unreliable. Journaling is particularly useful for protection against power failures, especially if your replica set resides in a single data center or power circuit. When a replica set runs with journaling, mongod (page 1021) instances can safely restart without any administrator intervention. Note: Journaling requires some resource overhead for write operations. Journaling has no effect on read performance, however. Journaling is enabled by default on all 64-bit builds of MongoDB v2.0 and greater.
59.10 Are write operations durable if write concern does not acknowledge writes?
Yes. However, if you want conrmation that a given write has arrived at the server, use write concern (page 398). The getLastError (page 831) command provides the facility for write concern. However, after the default write concern change (page 1185), the default write concern acknowledges all write operations, and unacknowledged writes must be explicitly congured. See the MongoDB Drivers and Client Libraries (page 555) documentation for your driver for more information.
59.11 How many arbiters do replica sets need?

Some congurations do not require any arbiter instances. Arbiters vote in elections for primary but do not replicate the data like secondary members.
59.7. What is the preferred replication method: master/slave or replica sets?
741
Replica sets require a majority of the remaining nodes present to elect a primary. Arbiters allow you to construct this majority without the overhead of adding replicating nodes to the system. There are many possible replica set architectures (page 395). If you have a three node replica set, you dont need an arbiter. But a common conguration consists of two replicating nodes, one of which is primary and the other is secondary, as well as an arbiter for the third node. This conguration makes it possible for the set to elect a primary in the event of a failure without requiring three replicating nodes. You may also consider adding an arbiter to a set if it has an equal number of nodes in two facilities and network partitions between the facilities are possible. In these cases, the arbiter will break the tie between the two facilities and allow the set to elect a new primary. See also: Replica Set Architectures and Deployment Patterns (page 395)
59.12 What information do arbiters exchange with the rest of the replica set?
Arbiters never receive the contents of a collection but do exchange the following data with the rest of the replica set: Credentials used to authenticate the arbiter with the replica set. All MongoDB processes within a replica set use keyles. These exchanges are encrypted. Replica set conguration data and voting data. This information is not encrypted. Only credential exchanges are encrypted. If your MongoDB deployment uses SSL, then all communications between arbiters and the other members of the replica set are secure. See the documentation for Connect to MongoDB with SSL (page 75) for more information. Run all arbiters on secure networks, as with all MongoDB components. See also: The overview of Arbiter Members of Replica Sets (page 388).
59.13 Which members of a replica set vote in elections?

All members of a replica set, unless the value of votes (page 465) is equal to 0, vote in elections. This includes all delayed (page 388), hidden (page 388) and secondary-only (page 388) members, as well as the arbiters (page 388). See also: Elections (page 389)
59.14 Do hidden members vote in replica set elections?

Hidden members (page 388) of term:replica :sets do vote in elections. To exclude a member from voting in an :election, change the value of the members votes (page 465) conguration to 0. See also: Elections (page 389)
742
59.15 Is it normal for replica set members to use different amounts of disk space?
Yes. Factors including: different oplog sizes, different levels of storage fragmentation, and MongoDBs data le preallocation can lead to some variation in storage utilization between nodes. Storage use disparities will be most pronounced when you add members at different times.
59.15. Is it normal for replica set members to use different amounts of disk space?
743
744
CHAPTER 60
FAQ: MongoDB Storage
Frequently Asked Questions: What are memory mapped les? (page 745) How do memory mapped les work? (page 745) How does MongoDB work with memory mapped les? (page 746) What are page faults? (page 746) What is the difference between soft and hard page faults? (page 746) What tools can I use to investigate storage use in MongoDB? (page 746) What is the working set? (page 746) Why are the les in my data directory larger than the data in my database? (page 747) How can I check the size of a collection? (page 748) How can I check the size of indexes? (page 748) How do I know when the server runs out of disk space? (page 748)
This document addresses common questions regarding MongoDBs storage system. If you dont nd the answer youre looking for, check the complete list of FAQs (page 709) or post your question to the MongoDB User Mailing List.
60.1 What are memory mapped les?

A memory-mapped le is a le with data that the operating system places in memory by way of the mmap() system call. mmap() thus maps the le to a region of virtual memory. Memory-mapped les are the critical piece of the storage engine in MongoDB. By using memory mapped les MongoDB can treat the contents of its data les as if they were in memory. This provides MongoDB with an extremely fast and simple method for accessing and manipulating data.
60.2 How do memory mapped les work?

Memory mapping assigns les to a block of virtual memory with a direct byte-for-byte correlation. Once mapped, the relationship between le and memory allows MongoDB to interact with the data in the le as if it were memory.
745
60.3 How does MongoDB work with memory mapped les?

MongoDB uses memory mapped les for managing and interacting with all data. MongoDB memory maps data les to memory as it accesses documents. Data that isnt accessed is not mapped to memory.
60.4 What are page faults?

Page faults will occur if youre attempting to access part of a memory-mapped le that isnt in memory. If there is free memory, then the operating system can nd the page on disk and load it to memory directly. However, if there is no free memory, the operating system must: nd a page in memory that is stale or no longer needed, and write the page to disk. read the requested page from disk and load it into memory. This process, particularly on an active system can take a long time, particularly in comparison to reading a page that is already in memory.
60.5 What is the difference between soft and hard page faults?
Page faults occur when MongoDB needs access to data that isnt currently in active memory. A hard page fault refers to situations when MongoDB must access a disk to access the data. A soft page fault, by contrast, merely moves memory pages from one list to another, such as from an operating system le cache. In production, MongoDB will rarely encounter soft page faults.
60.6 What tools can I use to investigate storage use in MongoDB?

The db.stats() (page 986) method in the mongo (page 1036) shell, returns the current state of the active database. The dbStats command (page 877) document describes the elds in the db.stats() (page 986) output.
60.7 What is the working set?

Working set represents the total body of data that the application uses in the course of normal operation. Often this is a subset of the total data size, but the specic size of the working set depends on actual moment-to-moment use of the database. If you run a query that requires MongoDB to scan every document in a collection, the working set will expand to include every document. Depending on physical memory size, this may cause documents in the working set to page out, or removed from physical memory by the operating system. The next time MongoDB needs to access these documents, MongoDB may incur a hard page fault. If you run a query that requires MongoDB to scan every document in a collection, the working set includes every active document in memory. For best performance, the majority of your active set should t in RAM.
746
Chapter 60. FAQ: MongoDB Storage
60.8 Why are the les in my data directory larger than the data in my database?
The data les in your data directory, which is the http://docs.mongodb.org/manual/data/db directory in default congurations, might be larger than the data set inserted into the database. Consider the following possible causes: Preallocated data les. In the data directory, MongoDB preallocates data les to a particular size, in part to prevent le system fragmentation. MongoDB names the rst data le <databasename>.0, the next <databasename>.1, etc. The rst le mongod (page 1021) allocates is 64 megabytes, the next 128 megabytes, and so on, up to 2 gigabytes, at which point all subsequent les are 2 gigabytes. The data les include les with allocated space but that hold no data. mongod (page 1021) may allocate a 1 gigabyte data le that may be 90% empty. For most larger databases, unused allocated space is small compared to the database. On Unix-like systems, mongod (page 1021) preallocates an additional data le and initializes the disk space to 0. Preallocating data les in the background prevents signicant delays when a new database le is next allocated. You can disable preallocation with the noprealloc (page 1084) run time option. However noprealloc (page 1084) is not intended for use in production environments: only use noprealloc (page 1084) for testing and with small data sets where you frequently drop databases. On Linux systems you can use hdparm to get an idea of how costly allocation might be:
time hdparm --fallocate $((1024*1024)) testfile
The oplog. If this mongod (page 1021) is a member of a replica set, the data directory includes the oplog.rs le, which is a preallocated capped collection in the local database. The default allocation is approximately 5% of disk space on 64-bit installations, see Oplog Sizing (page 392) for more information. In most cases, you should not need to resize the oplog. However, if you do, see Change the Size of the Oplog (page 435). The journal. The data directory contains the journal les, which store write operations on disk prior to MongoDB applying them to databases. See Journaling (page 71). Empty records. MongoDB maintains lists of empty records in data les when deleting documents and collections. MongoDB can reuse this space, but will never return this space to the operating system. To reclaim deleted space, use either of the following: compact (page 859), which defragments deleted space. compact (page 859) requires up to 2 gigabytes of extra disk space to run. Do not use compact (page 859) if you are critically low on disk space. repairDatabase (page 868), which rebuilds the database. Both options require additional disk space to run. For details, see Recover MongoDB Data following Unexpected Shutdown (page 52). Warning: repairDatabase (page 868) requires enough free disk space to hold both the old and new database les while the repair is running. Be aware that repairDatabase (page 868) will block all other operations and may take a long time to complete.
60.8. Why are the les in my data directory larger than the data in my database?
747
60.9 How can I check the size of a collection?

To view the size of a collection and other information, use the stats() (page 947) method from the mongo (page 1036) shell. The following example issues stats() (page 947) for the orders collection:
db.orders.stats();
To view specic measures of size, use these methods: db.collection.dataSize() (page 920): data size for the collection. db.collection.storageSize() (page 947): allocation size, including unused space. db.collection.totalSize() (page 948): the data size plus the index size. db.collection.totalIndexSize(): the index size. Also, the following scripts print the statistics for each database and collection:
db._adminCommand("listDatabases").databases.forEach(function (d) {mdb = db.getSiblingDB(d.name); prin
db._adminCommand("listDatabases").databases.forEach(function (d) {mdb = db.getSiblingDB(d.name); mdb.
60.10 How can I check the size of indexes?

To view the size of the data allocated for an index, use one of the following procedures in the mongo (page 1036) shell: Use the stats() (page 947) method using the index namespace. To retrieve a list of namespaces, issue the following command:
db.system.namespaces.find()
Check the value of indexSizes (page 874) in the output of the db.collection.stats() (page 947) command. Example Issue the following command to retrieve index namespaces:
db.system.namespaces.find()
The command returns a list similar to the following:

{"name" : "test.orders"} {"name" : "test.system.indexes"} {"name" : "test.orders.$_id_"}
View the size of the data allocated for the orders.$_id_ index with the following sequence of operations:
use test db.orders.$_id_.stats().indexSizes
60.11 How do I know when the server runs out of disk space?
If your server runs out of disk space for data les, you will see something like this in the log:
748
Thu Thu Thu Thu Thu Thu
Aug Aug Aug Aug Aug Aug
11 11 11 11 11 11
13:06:09 13:06:09 13:06:09 13:06:19 13:06:19 13:06:19
[FileAllocator] [FileAllocator] [FileAllocator] [FileAllocator] [FileAllocator] [FileAllocator]
allocating new data file error failed to allocate will try again in 10 allocating new data file error failed to allocate will try again in 10
dbms/test.13, filling with zeroes... new file: dbms/test.13 size: 2146435072 seconds dbms/test.13, filling with zeroes... new file: dbms/test.13 size: 2146435072 seconds
The server remains in this state forever, blocking all writes including deletes. However, reads still work. To delete some data and compact, using the compact (page 859) command, you must restart the server rst. If your server runs out of disk space for journal les, the server process will exit. By default, mongod (page 1021) creates journal les in a sub-directory of dbpath (page 1081) named journal. You may elect to put the journal les on another storage device using a lesystem mount or a symlink. Note: If you place the journal les on a separate storage device you will not be able to use a le system snapshot tool to capture a consistent snapshot of your data les and journal les.
60.11. How do I know when the server runs out of disk space?
749
750
CHAPTER 61
FAQ: Indexes
Frequently Asked Questions: Should you run ensureIndex() after every insert? (page 751) How do you know what indexes exist in a collection? (page 751) How do you determine the size of an index? (page 752) What happens if an index does not t into RAM? (page 752) How do you know what index a query used? (page 752) How do you determine what elds to index? (page 752) How do write operations affect indexes? (page 752) Will building a large index affect database performance? (page 752) Can I use index keys to constrain query matches? (page 753) Using $ne and $nin in a query is slow. Why? (page 753) Can I use a multi-key index to support a query for a whole array? (page 753) How can I effectively use indexes strategy for attribute lookups? (page 753)
This document addresses common questions regarding MongoDB indexes. If you dont nd the answer youre looking for, check the complete list of FAQs (page 709) or post your question to the MongoDB User Mailing List. See also Indexing Strategies (page 341).
61.1 Should you run ensureIndex() after every insert?

No. You only need to create an index once for a single collection. After initial creation, MongoDB automatically updates the index as data changes. While running ensureIndex() (page 921) is usually ok, if an index doesnt exist because of ongoing administrative work, a call to ensureIndex() (page 921) may disrupt database availability. Running ensureIndex() (page 921) can render a replica set inaccessible as the index creation is happening. See Build Indexes on Replica Sets (page 351).
61.2 How do you know what indexes exist in a collection?

To list a collections indexes, use the db.collection.getIndexes() (page 930) method or a similar method for your driver.
751
61.3 How do you determine the size of an index?

To check the sizes of the indexes on a collection, use db.collection.stats() (page 947).
61.4 What happens if an index does not t into RAM?

When an index is too large to t into RAM, MongoDB must read the index from disk, which is a much slower operation than reading from RAM. Keep in mind an index ts into RAM when your server has RAM available for the index combined with the rest of the working set. In certain cases, an index does not need to t entirely into RAM. For details, see Indexes that Hold Only Recent Values in RAM (page 345).
61.5 How do you know what index a query used?

To inspect how MongoDB processes a query, use the explain() (page 953) method in the mongo (page 1036) shell, or in your application driver.
61.6 How do you determine what elds to index?

A number of factors determine what elds to index, including selectivity (page 345), tting indexes into RAM, reusing indexes in multiple queries when possible, and creating indexes that can support all the elds in a given query. For detailed documentation on choosing which elds to index, see Indexing Strategies (page 341).
61.7 How do write operations affect indexes?

Any write operation that alters an indexed eld requires an update to the index in addition to the document itself. If you update a document that causes the document to grow beyond the allotted record size, then MongoDB must update all indexes that include this document as part of the update operation. Therefore, if your application is write-heavy, creating too many indexes might affect performance.
61.8 Will building a large index affect database performance?

Building an index can be an IO-intensive operation, especially if you have a large collection. This is true on any database system that supports secondary indexes, including MySQL. If you need to build an index on a large collection, consider building the index in the background. See Index Creation Options (page 335). If you build a large index without the background option, and if doing so causes the database to stop responding, do one of the following: Wait for the index to nish building. Kill the current operation (see db.killOp() (page 981)). The partial index will be deleted.
752
Chapter 61. FAQ: Indexes
61.9 Can I use index keys to constrain query matches?

You can use the min() (page 961) and max() (page 959) methods to constrain the results of the cursor returned from find() (page 924) by using index keys.
61.10 Using $ne and $nin in a query is slow. Why?

The $ne (page 767) and $nin (page 767) operators are not selective. See Create Queries that Ensure Selectivity (page 345). If you need to use these, it is often best to make sure that an additional, more selective criterion is part of the query.
61.11 Can I use a multi-key index to support a query for a whole array?
Not entirely. The index can partially support these queries because it can speed the selection of the rst element of the array; however, comparing all subsequent items in the array cannot use the index and must scan the documents individually.
61.12 How can I effectively use indexes strategy for attribute lookups?
For simple attribute lookups that dont require sorted result sets or range queries, consider creating a eld that contains an array of documents where each document has a eld (e.g. attrib ) that holds a specic type of attribute. You can index this attrib eld. For example, the attrib eld in the following document allows you to add an unlimited number of attributes types:
{ _id : ObjectId(...), attrib : [ { k: "color", { k: "shape": { k: "color": { k: "avail": ] }
v: v: v: v:
"red" }, "rectangle" }, "blue" }, true }
Both of the following queries could use the same { "attrib.k":
1, "attrib.v":
1 } index:
db.mycollection.find( { attrib: { $elemMatch : { k: "color", v: "blue" } } } ) db.mycollection.find( { attrib: { $elemMatch : { k: "avail", v: true } } } )
61.9. Can I use index keys to constrain query matches?
753
754
Chapter 61. FAQ: Indexes
CHAPTER 62
FAQ: MongoDB Diagnostics
Frequently Asked Questions: Where can I nd information about a mongod process that stopped running unexpectedly? (page 755) Does TCP keepalive time affect sharded clusters and replica sets? (page 756) Memory Diagnostics (page 756) Do I need to congure swap space? (page 756) Must my working set size t RAM? (page 757) How do I calculate how much RAM I need for my application? (page 757) How do I read memory statistics in the UNIX top command (page 757) Sharded Cluster Diagnostics (page 757) In a new sharded cluster, why does all data remains on one shard? (page 758) Why would one shard receive a disproportion amount of trafc in a sharded cluster? (page 758) What can prevent a sharded cluster from balancing? (page 758) Why do chunk migrations affect sharded cluster performance? (page 759) This document provides answers to common diagnostic questions and issues. If you dont nd the answer youre looking for, check the complete list of FAQs (page 709) or post your question to the MongoDB User Mailing List.
62.1 Where can I nd information about a mongod process that stopped running unexpectedly?
If mongod (page 1021) shuts down unexpectedly on a UNIX or UNIX-based platform, and if mongod (page 1021) fails to log a shutdown or error message, then check your system logs for messages pertaining to MongoDB. For example, for logs located in http://docs.mongodb.org/manual/var/log/messages, use the following commands:
sudo grep mongod /var/log/messages sudo grep score /var/log/messages
755
62.2 Does TCP keepalive time affect sharded clusters and replica sets?
If you experience socket errors between members of a sharded cluster or replica set, that do not have other reasonable causes, check the TCP keep alive value, which Linux systems store as the tcp_keepalive_time value. A common keep alive period is 7200 seconds (2 hours); however, different distributions and OS X may have different settings. For MongoDB, you will have better experiences with shorter keepalive periods, on the order of 300 seconds (ve minutes). On Linux systems you can use the following operation to check the value of tcp_keepalive_time:
cat /proc/sys/net/ipv4/tcp_keepalive_time
You can change the tcp_keepalive_time value with the following operation:
echo 300 > /proc/sys/net/ipv4/tcp_keepalive_time
The new tcp_keepalive_time value takes effect without requiring you to restart the mongod (page 1021) or mongos (page 1032) servers. When you reboot or restart your system you will need to set the new tcp_keepalive_time value, or see your operating systems documentation for setting the TCP keepalive value persistently. For OS X systems, issue the following command to view the keep alive setting:
sysctl net.inet.tcp.keepinit
To set a shorter keep alive period use the following invocation:

sysctl -w net.inet.tcp.keepinit=300
If your replica set or sharded cluster experiences keepalive-related issues, you must must alter the tcp_keepalive_time value on all machines hosting MongoDB processes. This includes all machines hosting mongos (page 1032) or mongod (page 1021) servers. Windows users should consider the Windows Server Technet Article on KeepAliveTime conguration for more information on setting keep alive for MongoDB deployments on Windows systems.
62.3 Memory Diagnostics

62.3.1 Do I need to congure swap space?
Always congure systems to have swap space. Without swap, your system may not be reliant in some situations with extreme memory constraints, memory leaks, or multiple programs using the same memory. Think of the swap space as something like a steam release valve that allows the system to release extra pressure without affecting the overall functioning of the system. Nevertheless, systems running MongoDB do not need swap for routine operation. Database les are memory-mapped (page 745) and should constitute most of your MongoDB memory use. Therefore, it is unlikely that mongod (page 1021) will ever use any swap space in normal operation. The operating system will release memory from the memory mapped les without needing swap and MongoDB can write data to the data les without needing the swap system.
756
Chapter 62. FAQ: MongoDB Diagnostics
62.3.2 Must my working set size t RAM?

Your working set should stay in memory to achieve good performance. Otherwise many random disk IOs will occur, and unless you are using SSD, this can be quite slow. One area to watch specically in managing the size of your working set is index access patterns. If you are inserting into indexes at random locations (as would happen with ids that are randomly generated by hashes), you will continually be updating the whole index. If instead you are able to create your ids in approximately ascending order (for example, day concatenated with a random id), all the updates will occur at the right side of the b-tree and the working set size for index pages will be much smaller. It is ne if databases and thus virtual size are much larger than RAM.
62.3.3 How do I calculate how much RAM I need for my application?

The amount of RAM you need depends on several factors, including but not limited to: The relationship between database storage (page 745) and working set. The operating systems cache strategy for LRU (Least Recently Used) The impact of journaling (page 71) The number or rate of page faults and other MMS gauges to detect when you need more RAM MongoDB defers to the operating system when loading data into memory from disk. It simply memory maps (page 745) all its data les and relies on the operating system to cache data. The OS typically evicts the leastrecently-used data from RAM when it runs low on memory. For example if clients access indexes more frequently than documents, then indexes will more likely stay in RAM, but it depends on your particular usage. To calculate how much RAM you need, you must calculate your working set size, or the portion of your data that clients use most often. This depends on your access patterns, what indexes you have, and the size of your documents. If page faults are infrequent, your working set ts in RAM. If fault rates rise higher than that, you risk performance degradation. This is less critical with SSD drives than with spinning disks.
62.3.4 How do I read memory statistics in the UNIX top command

Because mongod (page 1021) uses memory-mapped les (page 745), the memory statistics in top require interpretation in a special way. On a large database, VSIZE (virtual bytes) tends to be the size of the entire database. If the mongod (page 1021) doesnt have other processes running, RSIZE (resident bytes) is the total memory of the machine, as this counts le system cache contents. For Linux systems, use the vmstat command to help determine how the system uses memory. On OS X systems use vm_stat.
62.4 Sharded Cluster Diagnostics

The two most important factors in maintaining a successful sharded cluster are: choosing an appropriate shard key (page 493) and sufcient capacity to support current and future operations (page 487). You can prevent most issues encountered with sharding by ensuring that you choose the best possible shard key for your deployment and ensure that you are always adding additional capacity to your cluster well before the current resources become saturated. Continue reading for specic issues you may encounter in a production environment.
62.4. Sharded Cluster Diagnostics
757
62.4.1 In a new sharded cluster, why does all data remains on one shard?
Your cluster must have sufcient data for sharding to make sense. Sharding works by migrating chunks between the shards until each shard has roughly the same number of chunks. The default chunk size is 64 megabytes. MongoDB will not begin migrations until the imbalance of chunks in the cluster exceeds the migration threshold (page 497). While the default chunk size is congurable with the chunkSize (page 1090) setting, these behaviors help prevent unnecessary chunk migrations, which can degrade the performance of your cluster as a whole. If you have just deployed a sharded cluster, make sure that you have enough data to make sharding effective. If you do not have sufcient data to create more than eight 64 megabyte chunks, then all data will remain on one shard. Either lower the chunk size (page 498) setting, or add more data to the cluster. As a related problem, the system will split chunks only on inserts or updates, which means that if you congure sharding and do not continue to issue insert and update operations, the database will not create any chunks. You can either wait until your application inserts data or split chunks manually (page 521). Finally, if your shard key has a low cardinality (page 493), MongoDB may not be able to create sufcient splits among the data.
62.4.2 Why would one shard receive a disproportion amount of trafc in a sharded cluster?
In some situations, a single shard or a subset of the cluster will receive a disproportionate portion of the trafc and workload. In almost all cases this is the result of a shard key that does not effectively allow write scaling (page 494). Its also possible that you have hot chunks. In this case, you may be able to solve the problem by splitting and then migrating parts of these chunks. In the worst case, you may have to consider re-sharding your data and choosing a different shard key (page 495) to correct this pattern.
62.4.3 What can prevent a sharded cluster from balancing?

If you have just deployed your sharded cluster, you may want to consider the troubleshooting suggestions for a new cluster where data remains on a single shard (page 758). If the cluster was initially balanced, but later developed an uneven distribution of data, consider the following possible causes: You have deleted or removed a signicant amount of data from the cluster. If you have added additional data, it may have a different distribution with regards to its shard key. Your shard key has low cardinality (page 493) and MongoDB cannot split the chunks any further. Your data set is growing faster than the balancer can distribute data around the cluster. This is uncommon and typically is the result of: a balancing window (page 528) that is too short, given the rate of data growth. an uneven distribution of write operations (page 494) that requires more data migration. You may have to choose a different shard key to resolve this issue. poor network connectivity between shards, which may lead to chunk migrations that take too long to complete. Investigate your network conguration and interconnections between shards.
758
62.4.4 Why do chunk migrations affect sharded cluster performance?

If migrations impact your cluster or applications performance, consider the following options, depending on the nature of the impact: 1. If migrations only interrupt your clusters sporadically, you can limit the balancing window (page 528) to prevent balancing activity during peak hours. Ensure that there is enough time remaining to keep the data from becoming out of balance again. 2. If the balancer is always migrating chunks to the detriment of overall cluster performance: You may want to attempt decreasing the chunk size (page 523) to limit the size of the migration. Your cluster may be over capacity, and you may want to attempt to add one or two shards (page 510) to the cluster to distribute load. Its also possible that your shard key causes your application to direct all writes to a single shard. This kind of activity pattern can require the balancer to migrate most data soon after writing it. Consider redeploying your cluster with a shard key that provides better write scaling (page 494).
62.4. Sharded Cluster Diagnostics
759
760
Part XIV
Reference
761
CHAPTER 63
MongoDB Interface
63.1 Query, Update and Projection Operators

Query Selectors (page 763) Comparison (page 763) Logical (page 768) Element (page 772) JavaScript (page 774) Geospatial (page 777) Array (page 783) Update Operators (page 784) Fields (page 784) Array (page 788) Bitwise (page 795) Isolation (page 796) Projection Operators (page 796) Projection Operators (page 801)
63.1.1 Query Selectors

Comparison
Comparison Query Operators
Name $all (page 763) $gt (page 764) $gte (page 765) $in (page 765) $lt (page 766) $lte (page 766) $ne (page 767) $nin (page 767) $all
Description Matches arrays that contain all elements specied in the query. Matches values that are greater than the value specied in the query. Matches values that are equal to or greater than the value specied in the query. Matches any of the values that exist in an array specied in the query. Matches vales that are less than the value specied in the query. Matches values that are less than or equal to the value specied in the query. Matches all values that are not equal to the value specied in the query. Matches values that do not exist in an array specied to the query.
763
$all Syntax: { field: { $all: [ <value> , <value1> ... ] } $all (page 763) selects the documents where the field holds an array and contains all elements (e.g. <value>, <value1>, etc.) in the array. Consider the following example:
db.inventory.find( { tags: { $all: [ "appliances", "school", "book" ] } } )
This query selects all documents in the inventory collection where the tags eld contains an array with the elements, appliances, school, and book. Therefore, the above query will match documents in the inventory collection that have a tags eld that hold either of the following arrays:
[ "school", "book", "bag", "headphone", "appliances" ] [ "appliances", "school", "book" ]
The $all (page 763) operator exists to describe and specify arrays in MongoDB queries. However, you may use the $all (page 763) operator to select against a non-array field, as in the following example:
db.inventory.find( { qty: { $all: [ 50 ] } } )
However, use the following form to express the same query:

db.inventory.find( { qty: 50 } )
Both queries will select all documents in the inventory collection where the value of the qty eld equals 50. Note: In most cases, MongoDB does not treat arrays as sets. This operator provides a notable exception to this approach. In the current release queries that use the $all (page 763) operator must scan all the documents that match the rst element in the query array. As a result, even with an index to support the query, the operation may be long running, particularly when the rst element in the array is not very selective. See also: find() (page 924), update() (page 948), and $set (page 788). $gt $gt Syntax: {field: {$gt: value} } $gt (page 764) selects those documents where the value of the field is greater than (i.e. >) the specied value. Consider the following example:
db.inventory.find( { qty: { $gt: 20 } } )
This query will select all documents in the inventory collection where the qty eld value is greater than 20. Consider the following example which uses the $gt (page 764) operator with a eld from an embedded document:
764
Chapter 63. MongoDB Interface
db.inventory.update( { "carrier.fee": { $gt: 2 } }, { $set: { price: 9.99 } } )
This update() (page 948) operation will set the value of the price eld in the documents that contain the embedded document carrier whose fee eld value is greater than 2. See also: find() (page 924), update() (page 948), $set (page 788). $gte $gte Syntax: {field: {$gte: value} } $gte (page 765) selects the documents where the value of the field is greater than or equal to (i.e. >=) a specied value (e.g. value.) Consider the following example:
db.inventory.find( { qty: { $gte: 20 } } )
This query would select all documents in inventory where the qty eld value is greater than or equal to 20. Consider the following example which uses the $gte (page 765) operator with a eld from an embedded document:
db.inventory.update( { "carrier.fee": { $gte: 2 } }, { $set: { price: 9.99 } } )
This update() (page 948) operation will set the value of the price eld that contain the embedded document carrier whose fee eld value is greater than or equal to 2. See also: find() (page 924), update() (page 948), $set (page 788). $in $in Syntax: { field: { $in: [<value1>, <value2>, ... <valueN> ] } } $in (page 765) selects the documents where the field value equals any value in the specied array (e.g. <value1>, <value2>, etc.) Consider the following example:
db.inventory.find( { qty: { $in: [ 5, 15 ] } } )
This query selects all documents in the inventory collection where the qty eld value is either 5 or 15. Although you can express this query using the $or (page 770) operator, choose the $in (page 765) operator rather than the $or (page 770) operator when performing equality checks on the same eld. If the field holds an array, then the $in (page 765) operator selects the documents whose field holds an array that contains at least one element that matches a value in the specied array (e.g. <value1>, <value2>, etc.) Consider the following example:
db.inventory.update( { tags: { $in: ["appliances", "school"] } }, { $set: { sale:true } } )
63.1. Query, Update and Projection Operators
765
This update() (page 948) operation will set the sale eld value in the inventory collection where the tags eld holds an array with at least one element matching an element in the array ["appliances", "school"]. Note: When using two or more $in (page 765) expressions, the product of the number of distinct elements in the $in (page 765) arrays must be less than 4000000. Otherwise, MongoDB will throw an exception of "combinatorial limit of $in partitioning of result set exceeded". See also: find() (page 924), update() (page 948), $or (page 770), $set (page 788). $lt $lt Syntax: {field: {$lt: value} } $lt (page 766) selects the documents where the value of the field is less than (i.e. <) the specied value. Consider the following example:
db.inventory.find( { qty: { $lt: 20 } } )
This query will select all documents in the inventory collection where the qty eld value is less than 20. Consider the following example which uses the $lt (page 766) operator with a eld from an embedded document:
db.inventory.update( { "carrier.fee": { $lt: 20 } }, { $set: { price: 9.99 } } )
This update() (page 948) operation will set the price eld value in the documents that contain the embedded document carrier whose fee eld value is less than 20. See also: find() (page 924), update() (page 948), $set (page 788). $lte $lte Syntax: { field: { $lte: value} } $lte (page 766) selects the documents where the value of the field is less than or equal to (i.e. <=) the specied value. Consider the following example:
db.inventory.find( { qty: { $lte: 20 } } )
This query will select all documents in the inventory collection where the qty eld value is less than or equal to 20. Consider the following example which uses the $lt (page 766) operator with a eld from an embedded document:
db.inventory.update( { "carrier.fee": { $lte: 5 } }, { $set: { price: 9.99 } } )
This update() (page 948) operation will set the price eld value in the documents that contain the embedded document carrier whose fee eld value is less than or equal to 5. See also: find() (page 924), update() (page 948), $set (page 788). 766 Chapter 63. MongoDB Interface
$ne $ne Syntax: {field: {$ne: value} } $ne (page 767) selects the documents where the value of the field is not equal (i.e. !=) to the specied value. This includes documents that do not contain the field. Consider the following example:
db.inventory.find( { qty: { $ne: 20 } } )
This query will select all documents in the inventory collection where the qty eld value does not equal 20, including those documents that do not contain the qty eld. Consider the following example which uses the $ne (page 767) operator with a eld from an embedded document:
db.inventory.update( { "carrier.state": { $ne: "NY" } }, { $set: { qty: 20 } } )
This update() (page 948) operation will set the qty eld value in the documents that contains the embedded document carrier whose state eld value does not equal NY, or where the state eld or the carrier embedded document does not exist. See also: find() (page 924), update() (page 948), $set (page 788). $nin $nin Syntax: { field: { $nin: [ <value1>, <value2> ... <valueN> ]} } $nin (page 767) selects the documents where: the field value is not in the specied array or the field does not exist. Consider the following query:
db.inventory.find( { qty: { $nin: [ 5, 15 ] } } )
This query will select all documents in the inventory collection where the qty eld value does not equal 5 nor 15. The selected documents will include those documents that do not contain the qty eld. If the field holds an array, then the $nin (page 767) operator selects the documents whose field holds an array with no element equal to a value in the specied array (e.g. <value1>, <value2>, etc.). Consider the following query:
db.inventory.update( { tags: { $nin: [ "appliances", "school" ] } }, { $set: { sale: false } } )
This update() (page 948) operation will set the sale eld value in the inventory collection where the tags eld holds an array with no elements matching an element in the array ["appliances", "school"] or where a document does not contain the tags eld. See also: find() (page 924), update() (page 948), $set (page 788).
767
Logical
Logical Query Operators
Name $and (page 768) $nor (page 769) $not (page 770) $or (page 770)
Description Joins query clauses with a logical AND returns all documents that match the conditions of both clauses. Joins query clauses with a logical NOR returns all documents that fail to match both clauses. Inverts the effect of a query expression and returns documents that do not match the query expression. Joins query clauses with a logical OR returns all documents that match the conditions of either clause.
$and $and New in version 2.0. Syntax: { $and: <expressionN> } ] } [ { <expression1> }, { <expression2> } , ... , {
$and (page 768) performs a logical AND operation on an array of two or more expressions (e.g. <expression1>, <expression2>, etc.) and selects the documents that satisfy all the expressions in the array. The $and (page 768) operator uses short-circuit evaluation. If the rst expression (e.g. <expression1>) evaluates to false, MongoDB will not evaluate the remaining expressions. Consider the following example:
db.inventory.find({ $and: [ { price: 1.99 }, { qty: { $lt: 20 } }, { sale: true } ] } )
This query will select all documents in the inventory collection where: price eld value equals 1.99 and qty eld value is less than 20 and sale eld value is equal to true. MongoDB provides an implicit AND operation when specifying a comma separated list of expressions. For example, you may write the above query as:
db.inventory.find( { price: 1.99, qty: { $lt: 20 } , sale: true } )
If, however, a query requires an AND operation on the same eld such as { price: { $ne: 1.99 } } AND { price: { $exists: true } }, then either use the $and (page 768) operator for the two separate expressions or combine the operator expressions for the eld { price: { $ne: 1.99, $exists: true } }. Consider the following examples:
db.inventory.update( { $and: [ { price: { $ne: 1.99 } }, { price: { $exists: true } } ] }, { $se db.inventory.update( { price: { $ne: 1.99, $exists: true } } , { $set: { qty: 15 } } )
Both update() (page 948) operations will set the value of the qty eld in documents where: the price eld value does not equal 1.99 and the price eld exists.
768
See also: find() (page 924), update() (page 948), $ne (page 767), $exists (page 772), $set (page 788). $nor $nor Syntax: { $nor: <expressionN> } ] } [ { <expression1> }, { <expression2> }, ... {
$nor (page 769) performs a logical NOR operation on an array of two or more <expressions> and selects the documents that fail all the <expressions> in the array. Consider the following example:
db.inventory.find( { $nor: [ { price: 1.99 }, { qty: { $lt: 20 } }, { sale: true } ] } )
This query will select all documents in the inventory collection where: the price eld value does not equal 1.99 and the qty eld value is not less than 20 and the sale eld value is not equal to true including those documents that do not contain these eld(s). The exception in returning documents that do not contain the eld in the $nor (page 769) expression is when the $nor (page 769) operator is used with the $exists (page 772) operator. Consider the following query which uses only the $nor (page 769) operator:
db.inventory.find( { $nor: [ { price: 1.99 }, { sale: true } ] } )
This query will return all documents that: contain the price eld whose value is not equal to 1.99 and contain the sale eld whose value is not equal to true or contain the price eld whose value is not equal to 1.99 but do not contain the sale eld or do not contain the price eld but contain the sale eld whose value is not equal to true or do not contain the price eld and do not contain the sale eld Compare that with the following query which uses the $nor (page 769) operator with the $exists (page 772) operator:
db.inventory.find( { $nor: [ { price: 1.99 }, { price: { $exists: false } }, { sale: true }, { sale: { $exists: false } } ] } )
This query will return all documents that: contain the price eld whose value is not equal to 1.99 and contain the sale eld whose value is not equal to true See also: find() (page 924), update() (page 948), $set (page 788), $exists (page 772).
769
$not $not Syntax: { field: { $not: { <operator-expression> } } } $not (page 770) performs a logical NOT operation on the specied <operator-expression> and selects the documents that do not match the <operator-expression>. This includes documents that do not contain the field. Consider the following query:
db.inventory.find( { price: { $not: { $gt: 1.99 } } } )
This query will select all documents in the inventory collection where: the price eld value is less than or equal to 1.99 or the price eld does not exist { $not: { $gt: 1.99 } } is different from the $lte (page 766) operator. { $lt: returns only the documents where price eld exists and its value is less than or equal to 1.99. 1.99 }
Remember that the $not (page 770) operator only affects other operators and cannot check elds and documents independently. So, use the $not (page 770) operator for logical disjunctions and the $ne (page 767) operator to test the contents of elds directly. Consider the following behaviors when using the $not (page 770) operator: The operation of the $not (page 770) operator is consistent with the behavior of other operators but may yield unexpected results with some data types like arrays. The $not (page 770) operator does not support operations with the $regex (page 774) operator. Instead use http://docs.mongodb.org/manual// or in your driver interfaces, use your languages regular expression capability to create regular expression objects. Consider the following example which http://docs.mongodb.org/manual//: uses the pattern match expression
db.inventory.find( { item: { $not: /^p.*/ } } )
The query will select all documents in the inventory collection where the item eld value does not start with the letter p. If you are using Python, you can write the above query with the PyMongo driver and Pythons python:re.compile() method to compile a regular expression, as follows:
import re for noMatch in db.inventory.find( { "item": { "$not": re.compile("^p.*") } } ): print noMatch
See also: find() (page 924), update() (page 948), $set (page 788), $gt (page 764), $regex (page 774), PyMongo, driver. $or $or New in version 1.6. Changed in version 2.0: You may nest $or (page 770) operations; however, these expressions are not as efciently optimized as top-level.
770
Syntax: { $or: <expressionN> } ] }
[ { <expression1> }, { <expression2> }, ...
, {
The $or (page 770) operator performs a logical OR operation on an array of two or more <expressions> and selects the documents that satisfy at least one of the <expressions>. Consider the following query:
db.inventory.find( { price:1.99, $or: [ { qty: { $lt: 20 } }, { sale: true } ] } )
This query will select all documents in the inventory collection where: the price eld value equals 1.99 and either the qty eld value is less than 20 or the sale eld value is true. Consider the following example which uses the $or (page 770) operator to select elds from embedded documents:
db.inventory.update( { $or: [ { price:10.99 }, { "carrier.state": "NY"} ] }, { $set: { sale: tru
This update() (page 948) operation will set the value of the sale eld in the documents in the inventory collection where: the price eld value equals 10.99 or the carrier embedded document contains a eld state whose value equals NY. When using $or (page 770) with <expressions> that are equality checks for the value of the same eld, choose the $in (page 765) operator over the $or (page 770) operator. Consider the query to select all documents in the inventory collection where: either price eld value equals 1.99 or the sale eld value equals true, and either qty eld value equals 20 or qty eld value equals 50, The most effective query would be:
db.inventory.find ( { $or: [ { price: 1.99 }, { sale: true } ], qty: { $in: [20, 50] } } )
Consider the following behaviors when using the $or (page 770) operator: When using indexes with $or (page 770) queries, remember that each clause of an $or (page 770) query will execute in parallel. These clauses can each use their own index. Consider the following query:
db.inventory.find ( { $or: [ { price: 1.99 }, { sale: true } ] } )
For this query, you would create one index on price ( db.inventory.ensureIndex( { price: 1 } ) ) and another index on sale ( db.inventory.ensureIndex( { sale: 1 } ) ) rather than a compound index. Also, when using the $or (page 770) operator with the sort() (page 965) method in a query, the query will not use the indexes on the $or (page 770) elds. Consider the following query which adds a sort() (page 965) method to the above query:
db.inventory.find ( { $or: [ { price: 1.99 }, { sale: true } ] } ).sort({item:1})
This modied query will not use the index on price nor the index on sale. You cannot use the $or (page 770) with 2d geospatial queries (page 369). See also: find() (page 924), update() (page 948), $set (page 788), $and (page 768), sort() (page 965).
771
Element
Element Query Operators
Name $exists (page 772) $mod (page 772) $type (page 773)
Description Matches documents that have the specied eld. Performs a modulo operation on the value of a eld and selects documents with a specied result. Selects documents if a eld is of the specied type.
$exists $exists Syntax: { field:
{ $exists:
<boolean> } }
$exists (page 772) selects the documents that contain the eld if <boolean> is true. If <boolean> is false, the query only returns the documents that do not contain the eld. Documents that contain the eld but has the value null are not returned. MongoDB $exists does not correspond to SQL operator exists. For SQL exists, refer to the $in (page 765) operator. Consider the following example:
db.inventory.find( { qty: { $exists: true, $nin: [ 5, 15 ] } } )
This query will select all documents in the inventory collection where the qty eld exists and its value does not equal either 5 nor 15. See also: find() (page 924) $nin (page 767) $and (page 768) $in (page 765) How do I query for elds that have null values? (page 720) $mod $mod Syntax: { field: { $mod: [ divisor, remainder ]} } $mod (page 772) selects the documents where the field value divided by the divisor has the specied remainder. Consider the following example:
db.inventory.find( { qty: { $mod: [ 4, 0 ] } } )
This query will select all documents in the inventory collection where the qty eld value modulo 4 equals 0, such as documents with qty value equal to 0 or 12. In some cases, you can query using the $mod (page 772) operator rather than the more expensive $where (page 775) operator. Consider the following example using the $mod (page 772) operator:
772
db.inventory.find( { qty: { $mod: [ 4, 0 ] } } )
The above query is less expensive than the following query which uses the $where (page 775) operator:
db.inventory.find( { $where: "this.qty % 4 == 0" } )
See also: find() (page 924), update() (page 948), $set (page 788). $type $type Syntax: { field:
{ $type:
<BSON type> } }
$type (page 773) selects the documents where the value of the field is the specied BSON type. Consider the following example:
db.inventory.find( { price: { $type : 1 } } )
This query will select all documents in the inventory collection where the price eld value is a Double. If the field holds an array, the $type (page 773) operator performs the type check against the array elements and not the field. Consider the following example where the tags eld holds an array:
db.inventory.find( { tags: { $type : 4 } } )
This query will select all documents in the inventory collection where the tags array contains an element that is itself an array. If instead you want to determine whether the tags eld is an array type, use the $where (page 775) operator:
db.inventory.find( { $where : "Array.isArray(this.tags)" } )
See the SERVER-1475 for more information about the array type. Refer to the following table for the available BSON types and their corresponding numbers. Type Double String Object Array Binary data Object id Boolean Date Null Regular Expression JavaScript Symbol JavaScript (with scope) 32-bit integer Timestamp 64-bit integer Min key Max key Number 1 2 3 4 5 7 8 9 10 11 13 14 15 16 17 18 255 127 773
MinKey and MaxKey compare less than and greater than all other possible BSON element values, respectively, and exist primarily for internal use. Note: To query if a eld value is a MinKey, you must use the $type (page 773) with -1 as in the following example:
db.collection.find( { field: { $type: -1 } } )
Example Consider the following example operation sequence that demonstrates both type comparison and the special MinKey and MaxKey values:
db.test.insert( db.test.insert( db.test.insert( db.test.insert( db.test.insert( db.test.insert( {x {x {x {x {x {x : : : : : : 3}); 2.9} ); new Date()} ); true } ); MaxKey } ) MinKey } )
db.test.find().sort({x:1}) { "_id" : ObjectId("4b04094b7c65b846e2090112"), { "_id" : ObjectId("4b03155dce8de6586fb002c7"), { "_id" : ObjectId("4b03154cce8de6586fb002c6"), { "_id" : ObjectId("4b031566ce8de6586fb002c9"), { "_id" : ObjectId("4b031563ce8de6586fb002c8"), { "_id" : ObjectId("4b0409487c65b846e2090111"),
"x" "x" "x" "x" "x" "x"
: : : : : :
{ $minKey : 1 } } 2.9 } 3 } true } "Tue Jul 25 2012 18:42:03 GMT-0500 (EST)" { $maxKey : 1 } }
To query for the minimum value of a shard key of a sharded cluster, use the following operation when connected to the mongos (page 1032):
use config db.chunks.find( { "min.shardKey": { $type: -1 } } )
Warning: Storing values of the different types in the same eld in a collection is strongly discouraged. See also: find() (page 924), insert() (page 936), $where (page 775), BSON , shard key, sharded cluster . JavaScript
JavaScript Query Operators
Name $regex (page 774) $where (page 775)
Description Selects documents where values match a specied regular expression. Matches documents that satisfy a JavaScript expression.
$regex $regex The $regex (page 774) operator provides regular expression capabilities for pattern matching strings in queries. MongoDB uses Perl compatible regular expressions (i.e. PCRE.)
774
You can specify regular expressions using regular expression objects or using the $regex (page 774) operator. The following examples are equivalent:
db.collection.find( { field: /acme.*corp/i } ); db.collection.find( { field: { $regex: acme.*corp, $options: i } } );
These expressions match all documents in collection where the value of field matches the caseinsensitive regular expression acme.*corp. $regex (page 774) uses Perl Compatible Regular Expressions (PCRE) as the matching engine. $options $regex (page 774) provides four option ags: i toggles case insensitivity, and allows all letters in the pattern to match upper and lower cases. m toggles multiline regular expression. Without this option, all regular expression match within one line. If there are no newline characters (e.g. \n) or no start/end of line construct, the m option has no effect. x toggles an extended capability. When set, $regex (page 774) ignores all white space characters unless escaped or included in a character class. Additionally, it ignores characters between an un-escaped # character and the next new line, so that you may include comments in complicated patterns. This only applies to data characters; white space characters may never appear within special character sequences in a pattern. The x option does not affect the handling of the VT character (i.e. code 11.) New in version 1.9.0. s allows the dot (e.g. .) character to match all characters including newline characters. $regex (page 774) only provides the i and m options for the native JavaScript regular expression objects (e.g. http://docs.mongodb.org/manual/acme.*corp/i). To use x and s you must use the $regex (page 774) operator with the $options (page 775) syntax. To combine a regular expression match with other operators, you need to use the $regex (page 774) operator. For example:
db.collection.find( { field: { $regex: /acme.*corp/i, $nin: [ acmeblahcorp ] } );
This expression returns all instances of field in collection that match the case insensitive regular expression acme.*corp that dont match acmeblahcorp. $regex (page 774) can only use an index efciently when the regular expression has an anchor for the beginning (i.e. ^) of a string and is a case-sensitive match. Additionally, while http://docs.mongodb.org/manual/â/, http://docs.mongodb.org/manual/â.*/, and http://docs.mongodb.org/manual/â.*$/ match equivalent strings, they have different performance characteristics. All of these expressions use an index if an appropriate index exists; however, http://docs.mongodb.org/manual/â.*/, and http://docs.mongodb.org/manual/â.*$/ are slower. http://docs.mongodb.org/manual/â/ can stop scanning after matching the prex. $where $where Use the $where (page 775) operator to pass either a string containing a JavaScript expression or a full JavaScript function to the query system. The $where (page 775) provides greater exibility, but requires that the database processes the JavaScript expression or function for each document in the collection. Reference the document in the JavaScript expression or function using either this or obj .
775
Warning: Do not write to the database within the $where (page 775) JavaScript function. $where (page 775) evaluates JavaScript and cannot take advantage of indexes. Therefore, query performance improves when you express your query using the standard MongoDB operators (e.g., $gt (page 764), $in (page 765)). In general, you should use $where (page 775) only when you cant express your query using another operator. If you must use $where (page 775), try to include at least one other standard query operator to lter the result set. Using $where (page 775) alone requires a table scan. Consider the following examples:
db.myCollection.find( { $where: "this.credits == this.debits" } ); db.myCollection.find( { $where: "obj.credits == obj.debits" } ); db.myCollection.find( { $where: function() { return (this.credits == this.debits) } } ); db.myCollection.find( { $where: function() { return obj.credits == obj.debits; } } );
Additionally, if the query consists only of the $where (page 775) operator, you can pass in just the JavaScript expression or JavaScript functions, as in the following examples:
db.myCollection.find( "this.credits == this.debits || this.credits > this.debits" );
db.myCollection.find( function() { return (this.credits == this.debits || this.credits > this.de
You can include both the standard MongoDB operators and the $where (page 775) operator in your query, as in the following examples:
db.myCollection.find( { active: true, $where: "this.credits - this.debits < 0" } ); db.myCollection.find( { active: true, $where: function() { return obj.credits - obj.debits < 0;
Using normal non-$where (page 775) query statements provides the following performance advantages: MongoDB will evaluate non-$where (page 775) components of query before $where (page 775) statements. If the non-$where (page 775) statements match no documents, MongoDB will not perform any query evaluation using $where (page 775). The non-$where (page 775) query statements may use an index. Note: Changed in version 2.4. In MongoDB 2.4, map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions cannot access certain global functions or properties, such as db, that are available in the mongo (page 1036) shell. When upgrading to MongoDB 2.4, you will need to refactor your code if your map-reduce operations (page 814), group (page 810) commands, or $where (page 775) operator expressions include any global shell functions or properties that are no longer available, such as db. The following shell functions and properties are available to map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions in MongoDB 2.4:
776
Available Properties args MaxKey MinKey
Available Functions assert() BinData() DBPointer() DBRef() doassert() emit() gc() HexData() hex_md5() isNumber() isObject() ISODate() isString() Map() MD5() NumberInt() NumberLong() ObjectId() print() printjson() printjsononeline() sleep() Timestamp() tojson() tojsononeline() tojsonObject() UUID() version()
Geospatial
Geospatial Query Operators
Operators
Name $geoIntersects (page 777) $geoWithin (page 778) $nearSphere (page 779) $near (page 780)
Description Selects geometries that intersect with a GeoJSON geometry. Selects geometries within a bounding GeoJSON geometry. Returns geospatial objects in proximity to a point on a sphere. Returns geospatial objects in proximity to a point.
$geoIntersects $geoIntersects New in version 2.4. The $geoIntersects (page 777) operator is a geospatial query operator that selects all locations that intersect with a GeoJSON object. A location intersects a GeoJSON object if the intersection is non-empty. This includes documents that have a shared edge. The $geoIntersects (page 777) operator uses spherical geometry. The 2dsphere geospatial index supports $geoIntersects (page 777). To query for intersection, pass the GeoJSON object to $geoIntersects (page 777) through the $geometry operator. Use the following syntax:
db.<collection>.find( { <location field> : { $geoIntersects : { $geometry : { type : "<GeoJSON object type>" , coordinates : [ <coordinates> ] } } } } )
Important: Specify coordinates in this order: longitude, latitude.
777
The following example uses $geoIntersects (page 777) to select all indexed points and shapes that intersect with the polygon dened by the coordinates array.
db.places.find( { loc : { $geoIntersects : { $geometry : { type : "Polygon" , coordinates: [ [ [ 0 , 0 ] , [ 3 , 6 ] , [ 6 , 1 ] , [ 0 , 0 ] ] ] } } } } } )
Note: Any geometry specied with GeoJSON to $geoIntersects (page 777) queries, must t within a single hemisphere. MongoDB interprets geometries larger than half of the sphere as queries for the smaller of the complementary geometries.
$geoWithin $geoWithin New in version 2.4: $geoWithin (page 778) replaces $within (page 779) which is deprecated. The $geoWithin (page 778) operator is a geospatial query operator that queries for a dened point, line or shape that exists entirely within another dened shape. When determining inclusion, MongoDB considers the border of a shape to be part of the shape, subject to the precision of oating point numbers. The $geoWithin (page 778) operator queries for inclusion in a GeoJSON polygon or a shape dened by legacy coordinate pairs. The $geoWithin (page 778) operator does not return sorted results. As a result MongoDB can return $geoWithin (page 778) queries more quickly than geospatial $near (page 780) or $nearSphere (page 779) queries, which sort results. The 2dsphere and 2d indexes both support the $geoWithin (page 778) operator. Changed in version 2.2.3: $geoWithin (page 778) does not require a geospatial index. However, a geospatial index will improve query performance. If querying for inclusion in a GeoJSON polygon on a sphere, pass the polygon to $geoWithin (page 778) through the $geometry operator. Coordinates of a polygon are an array of LinearRing coordinate arrays. The rst element in the array represents the exterior ring. Any subsequent elements represent interior rings (or holes). For a polygon with only an exterior ring use following syntax:
db.<collection>.find( { <location field> : { $geoWithin : { $geometry : { type : "Polygon" , coordinates : [ [ [ <lng1>, <lat1> ] , [ <lng2>, <lat2> ] ... ] } } } } )
Important: Specify coordinates in this order: longitude, latitude. For a polygon with an exterior and interior ring use following syntax:
db.<collection>.find( { <location field> : { $geoWithin : { $geometry : { type : "Polygon" ,
778
coordinates : [ [ [ <lng1>, <lat1> ] , [ <lng2>, <lat2> ] ... ] [ [ <lngA>, <latA> ] , [ <lngB>, <latB> ] ... ] } } } } )
The following example selects all indexed points and shapes that exist entirely within a GeoJSON polygon:
db.places.find( { loc : { $geoWithin : { $geometry : { type : "Polygon" , coordinates: [ [ [ 0 , 0 ] , [ 3 , 6 ] , [ 6 , 1 ] , [ 0 , 0 ] ] ] } } } } )
If querying for inclusion in a shape dened by legacy coordinate pairs on a plane, use the following syntax:
db.<collection>.find( { <location field> : { $geoWithin : { <shape operator> : <coordinates> } } } )
For the syntax of a shape operator, see: $box $polygon $center (denes a circle) $centerSphere (denes a circle on a sphere) Note: Any geometry specied with GeoJSON to $geoWithin (page 778) queries, must t within a single hemisphere. MongoDB interprets geometries larger than half of the sphere as queries for the smaller of the complementary geometries. $within Deprecated since version 2.4: $geoWithin (page 778) replaces $within (page 779) in MongoDB 2.4. $nearSphere $nearSphere New in version 1.8. The $nearSphere (page 779) operator species a point for which a geospatial query returns the 100 closest documents, sorted from nearest to farthest. MongoDB calculates distances for $nearSphere (page 779) using spherical geometry. The $nearSphere (page 779) operator queries for points dened by either GeoJSON objects or legacy coordinate pairs. The optional $maxDistance operator limits a $nearSphere (page 779) query to return only those documents that fall within a maximum distance of a point. If you use $maxDistance on GeoJSON points, the distance is measured in meters. If you use $maxDistance on legacy coordinate pairs, the distance is measured in radians. The $nearSphere (page 779) operator requires a geospatial index. The 2dsphere and 2d indexes both support $nearSphere (page 779). In a 2dsphere index, a grid coordinate is interpreted as a GeoJSON point. Important: If you use longitude and latitude, specify longitude rst. For queries on GeoJSON data, use the following syntax: 63.1. Query, Update and Projection Operators 779
db.<collection>.find( { <location field> : { $nearSphere : { $geometry : { type : "Point" , coordinates : [ <longitude> , <latitude> ] } , $maxDistance : <distance in meters> } } } )
For queries on legacy coordinate pairs, use the following syntax:

db.<collection>.find( { <location field> : { $nearSphere: [ <x> , <y> ] , $maxDistance: <distance in radians> } } )
The following example selects the 100 documents with legacy coordinates pairs nearest to [ 40 , 5 ], as calculated by spherical geometry:
db.places.find( { loc : { $nearSphere : [ 40 , 5 ] $maxDistance : 10 } } )
$near $near Changed in version 2.4. The $near (page 780) operator species a point for which a geospatial query returns the 100 closest documents. The query sorts the documents from nearest to farthest. The $near (page 780) operator can query for a GeoJSON point or for a point dened by legacy coordinate pairs. The optional $maxDistance operator limits a $near (page 780) query to return only those documents that fall within a maximum distance of a point. If you query for a GeoJSON point, specify $maxDistance in meters. If you query for legacy coordinate pairs, specify $maxDistance in radians. The $near (page 780) operator requires a geospatial index. For GeoJSON points, use a 2dsphere index. For legacy coordinate pairs, use a 2d index. For queries on GeoJSON data, use the following syntax:
db.<collection>.find( { <location field> : { $near : { $geometry : { type : "Point" , coordinates : [ <longitude> , <latitude> ] } }, $maxDistance : <distance in meters> } } )
Important: Specify coordinates in this order: longitude, latitude. The following example selects the 100 documents with coordinates nearest to [ 40 , 5 ] and limits the maximum distance to 100 meters from the specied GeoJSON point:
db.places.find( { loc : { $near : { $geometry : { type : "Point" ,
780
coordinates: [ 40 , 5 ] } }, $maxDistance : 100 } } )
For queries on legacy coordinate pairs, use the following syntax:

db.<collection>.find( { <location field> : { $near : [ <x> , <y> ] , $maxDistance: <distance> } } )
Important: If you use longitude and latitude, specify longitude rst. The following example selects the 100 documents with coordinates nearest to [ 40 , 5 ]:
db.places.find( { loc : { $near : [ 40 , 5 ] , $maxDistance : 10 } } )
Note: You can further limit the number of results using cursor.limit() (page 959). Specifying a batch size (i.e. batchSize() (page 952)) in conjunction with queries that use the $near (page 780) is not dened. See SERVER-5236 for more information.
Geospatial Query Compatibility While numerous combinations of query operators are possible, the following table shows the recommended operators for different types of queries. The table uses the $geoWithin (page 778), $geoIntersects (page 777) and $near (page 780) operators.
781
Query Document Returns points, lines and polygons
Geometry of the Query Condition
Surface Type for Query Calculation
Units for Query Calculation
Supported by this Index
polygon { $geoWithin : { $geometry : <GeoJSON Polygon> } } point, line or polygon { $geoIntersects : { $geometry : <GeoJSON> } } point { $near : { $geometry : <GeoJSON Point>, $maxDistance : d } } Returns points only rectangle { $geoWithin : { $box : [[x1, y1], [x2, y2]] } } polygon { $geoWithin : { $polygon : [[x1, y1], [x1, y2], [x2, y2], [x2, y1]] } } circular region { $geoWithin : { $center : [x1, y1], $maxDistance : d } } circular region { $geoWithin : { $centerSphere : [[x, y], radius] } } point { $near : [x1, y1], $maxDistance : d }
sphere
meters
2dsphere
sphere
meters
2dsphere
sphere
meters
2dsphere The index is required.
at
at units
2d
at
at units
2d
at
at units
2d
sphere
radians
2d 2dsphere
at / at units
at units
2d The index is required.
782
Array
Query Operator Array
Name $elemMatch (page 783) $size (page 783) $elemMatch (query)
Description Selects documents if element in the array eld matches all the specied $elemMatch (page 783) condition. Selects documents if the array eld is a specied size. See also:
$elemMatch (projection) (page 798) $elemMatch New in version 1.4. The $elemMatch (page 783) operator matches more than one component within an array element. For example,
db.collection.find( { array: { $elemMatch: { value1: 1, value2: { $gt: 1 } } } } );
returns all documents in collection where the array array satises all of the conditions in the $elemMatch (page 783) expression, or where the value of value1 is 1 and the value of value2 is greater than 1. Matching arrays must have at least one element that matches all specied criteria. Therefore, the following document would not match the above query:
{ array: [ { value1:1, value2:0 }, { value1:2, value2:2 } ] }
while the following document would match this query:

{ array: [ { value1:1, value2:0 }, { value1:1, value2:2 } ] }
$size $size The $size (page 783) operator matches any array with the number of elements specied by the argument. For example:
db.collection.find( { field: { $size: 2 } } );
returns all documents in collection where field is an array with 2 elements. For instance, the above expression will return { field: [ red, green ] } and { field: [ apple, lime ] } but not { field: fruit } or { field: [ orange, lemon, grapefruit ] }. To match elds with only one element within an array use $size (page 783) with a value of 1, as follows:
db.collection.find( { field: { $size: 1 } } );
$size (page 783) does not accept ranges of values. To select documents based on elds with different numbers of elements, create a counter eld that you increment when you add elements to a eld. Queries cannot use indexes for the $size (page 783) portion of a query, although the other portions of a query can use indexes if applicable.
783
63.1.2 Update Operators

Fields
Field Update Operators
Name $inc (page 784) $rename (page 784) $setOnInsert (page 787) $set (page 788) $unset (page 788) $inc $inc
Description Increments the value of the eld by the specied amount. Renames a eld. Sets the value of a eld upon documentation creation during an upsert. Has no effect on update operations that modify existing documents. Sets the value of a eld in an existing document. Removes the specied eld from an existing document.
The $inc (page 784) operator increments a value of a eld by a specied amount. If the eld does not exist, $inc (page 784) sets the eld to the specied amount. $inc (page 784) accepts positive and negative incremental amounts. The following example increments the value of field1 by the value of amount for the rst matching document in the collection where field equals value:
db.collection.update( { field: value }, { $inc: { field1: amount } } );
To update all matching documents in the collection, specify multi:true in the update() (page 948) method:
db.collection.update( { age: 20 }, { $inc: { age: 1 } }, { multi: true } ); db.collection.update( { name: "John" }, { $inc: { age: 2 } }, { multi: true } );
The rst update() (page 948) operation increments the value of the age eld by 1 for all documents in the collection that have an age eld equal to 20. The second operation increments the value of the age eld by 2 for all documents in the collection with the name eld equal to "John". $rename $rename New in version 1.7.2. Syntax: {$rename: ... } } { <old name1>: <new name1>, <old name2>: <new name2>,
The $rename (page 784) operator updates the name of a eld. The new eld name must differ from the existing eld name. Consider the following example:
db.students.update( { _id: 1 }, { $rename: { nickname: alias, cell: mobile } } )
This operation renames the eld nickname to alias, and the eld cell to mobile. If the document already has a eld with the new eld name, the $rename (page 784) operator removes that eld and renames the eld with the old eld name to the new eld name.
784
The $rename (page 784) operator will expand arrays and sub-documents to nd a match for eld names. When renaming a eld in a sub-document to another sub-document or to a regular eld, the sub-document itself remains. Consider the following examples involving the sub-document of the following document:
{ "_id": 1, "alias": [ "The American Cincinnatus", "The American Fabius" ], "mobile": "555-555-5555", "nmae": { "first" : "george", "last" : "washington" } }
To rename a sub-document, call the $rename (page 784) operator with the name of the sub-document as you would any other eld:
db.students.update( { _id: 1 }, { $rename: { "nmae": "name" } } )
This operation renames the sub-document nmae to name:

{ "_id": 1, "alias": [ "The American Cincinnatus", "The American Fabius" ], "mobile": "555-555-5555", "name": { "first" : "george", "last" : "washington" } }
To rename a eld within a sub-document, call the $rename (page 784) operator using the dot notation (page 188) to refer to the eld. Include the name of the sub-document in the new eld name to ensure the eld remains in the sub-document:
db.students.update( { _id: 1 }, { $rename: { "name.first": "name.fname" } } )
This operation renames the sub-document eld first to fname:

{ "_id" : 1, "alias" : [ "The American Cincinnatus", "The American Fabius" ], "mobile" : "555-555-5555", "name" : { "fname" : "george", "last" : "washington" } }
To rename a eld within a sub-document and move it to another sub-document, call the $rename (page 784) operator using the dot notation (page 188) to refer to the eld. Include the name of the new sub-document in the new name:
db.students.update( { _id: 1 }, { $rename: { "name.last": "contact.lname" } } )
This operation renames the sub-document eld last to lname and moves it to the sub-document contact:
{ "_id" : 1, "alias" : [ "The American Cincinnatus", "The American Fabius" ], "contact" : { "lname" : "washington" }, "mobile" : "555-555-5555", "name" : { "fname" : "george" } }
If the new eld name does not include a sub-document name, the eld moves out of the subdocument and becomes a regular document eld. Consider the following behavior when the specied old eld name does not exist:
785
When renaming a single eld and the existing eld name refers to a non-existing eld, the $rename (page 784) operator does nothing, as in the following:
db.students.update( { _id: 1 }, { $rename: { wife: spouse } } )
This operation does nothing because there is no eld named wife. When renaming multiple elds and all of the old eld names refer to non-existing elds, the $rename (page 784) operator does nothing, as in the following:
db.students.update( { _id: 1 }, { $rename: { wife: spouse, vice: vp, office: term } } )
This operation does nothing because there are no elds named wife, vice, and office. When renaming multiple elds and some but not all old eld names refer to non-existing elds, the $rename (page 784) operator performs the following operations: Changed in version 2.2. Renames the elds that exist to the specied new eld names. Ignores the non-existing elds. Consider the following query that renames both an existing eld mobile and a non-existing eld wife. The eld named wife does not exist and $rename (page 784) sets the eld to a name that already exists alias.
db.students.update( { _id: 1 }, { $rename: { wife: alias, mobile: cell } } )
This operation renames the mobile eld to cell, and has no other impact action occurs.
{ "_id" : 1, "alias" : [ "The American Cincinnatus", "The American Fabius" ], "cell" : "555-555-5555", "name" : { "lname" : "washington" }, "places" : { "d" : "Mt Vernon", "b" : "Colonial Beach" } }
Note: Before version 2.2, when renaming multiple elds and only some (but not all) old eld names refer to non-existing elds: For the elds with the old names that do exist, the $rename (page 784) operator renames these elds to the specied new eld names. For the elds with the old names that do not exist: *if no eld exists with the new eld name, the $rename (page 784) operator does nothing. *if elds already exist with the new eld names, the $rename (page 784) operator drops these elds. Consider the following operation that renames both the eld mobile, which exists, and the eld wife, which does not exist. The operation tries to set the eld named wife to alias, which is the name of an existing eld:
db.students.update( { _id: 1 }, { $rename: { wife: alias, mobile: cell } } )
Before 2.2, the operation renames the eld mobile to cell and drops the alias eld even though the eld wife does not exist:
786
{ "_id" : 1, "cell" : "555-555-5555", "name" : { "lname" : "washington" }, "places" : { "d" : "Mt Vernon", "b" : "Colonial Beach" } }
$setOnInsert $setOnInsert New in version 2.4. The $setOnInsert (page 787) operator assigns values to elds during an upsert (page 948) only when using the upsert option to the update() (page 948) operation performs an insert.
db.collection.update( <query>, { $setOnInsert: { <field1>: <value1>, ... } }, { upsert: true } )
Example A collection named products contains no documents. Then, the following upsert (page 948) operation performs an insert and applies the $setOnInsert (page 787) to set the eld defaultQty to 100:
db.products.update( { _id: 1 }, { $setOnInsert: { defaultQty: 100 } }, { upsert: true } )
The products collection contains the newly-inserted document:

{ "_id" : 1, "defaultQty" : 100 }
Note: The $setOnInsert (page 787) operator only affects update() (page 948) operations with the upsert ag that perform an insert (page 201). If the update() (page 948) has the upsert ag and performs an update (page 219), $setOnInsert (page 787) has no effect. Example A collection named products has the following document:
{ "_id" : 1, "defaultQty" : 100 }
The following update() (page 948) with the upsert ag operation performs an update:
db.products.update( { _id: 1 }, { $setOnInsert: { defaultQty: 500, inStock: true }, $set: { item: "apple" } }, { upsert: true } )
787
Because the update() (page 948) with upsert operation only performs an update, MongoDB ignores the $setOnInsert (page 787) operation and only applies the $set (page 788) operation. The products collection now contains the following modied document:
{ "_id" : 1, "defaultQty" : 100, "item" : "apple" }
$set $set Use the $set (page 788) operator to set a particular value. The $set (page 788) operator requires the following syntax:
db.collection.update( { field: value1 }, { $set: { field1: value2 } } );
This statement updates in the document in collection where field matches value1 by replacing the value of the eld field1 with value2. This operator will add the specied eld or elds if they do not exist in this document or replace the existing value of the specied eld(s) if they already exist. $unset $unset The $unset (page 788) operator deletes a particular eld. Consider the following example:
db.collection.update( { field: value1 }, { $unset: { field1: "" } } );
The above example deletes field1 in collection from documents where field has a value of value1. The value of the eld in the $unset (page 788) statement (i.e. "" above) does not impact the operation. If documents match the initial query (e.g. { field: value1 } above) but do not have the eld specied in the $unset (page 788) operation (e.g. field1), then the statement has no effect on the document. Array
Array Update Operators
Name $ (page 788) $addToSet (page 789) $pop (page 790) $pullAll (page 790) $pull (page 791) $pushAll (page 791) $push (page 792)
Description Acts as a placeholder to update the rst element that matches the query condition in an update. Adds elements to an existing array only if they do not already exist in the set. Removes the rst or last item of an array. Removes multiple values from an array. Removes items from an array that match a query statement. Deprecated. Adds several items to an array. Adds an item to an array.
Update Operators
$ (query) $ Syntax: { "<array>.$" :
value }
788
The positional $ (page 788) operator identies an element in an array eld to update without explicitly specifying the position of the element in the array. To project, or return, an array element from a read operation, see the $ (page 797) projection operator. When used with the update() (page 948) method, the positional $ (page 788) operator acts as a placeholder for the rst element that matches the query document (page 190), and the array eld must appear as part of the query document.
db.collection.update( { <array>: value ... }, { <update operator>: { "<array>.$" : value } } )
Consider a collection students with the following documents:

{ "_id" : 1, "grades" : [ 80, 85, 90 ] } { "_id" : 2, "grades" : [ 88, 90, 92 ] } { "_id" : 3, "grades" : [ 85, 100, 90 ] }
To update 80 to 82 in the grades array in the rst document, use the positional $ (page 788) operator if you do not know the position of the element in the array:
db.students.update( { _id: 1, grades: 80 }, { $set: { "grades.$" : 82 } } )
Remember that the positional $ (page 788) operator acts as a placeholder for the rst match of the update query document (page 190). The positional $ (page 788) operator facilitates updates to arrays that contain embedded documents. Use the positional $ (page 788) operator to access the elds in the embedded documents with the dot notation (page 188) on the $ (page 788) operator.
db.collection.update( { <query selector> }, { <update operator>: { "array.$.field" : value } } )
Consider the following document in the students collection whose grades eld value is an array of embedded documents:
{ "_id" : 4, "grades" : [ { grade: 80, mean: 75, std: 8 }, { grade: 85, mean: 90, std: 5 }, { grade: 90, mean: 85, std: 3 } ] }
Use the positional $ (page 788) operator to update the value of the std eld in the embedded document with the grade of 85:
db.students.update( { _id: 4, "grades.grade": 85 }, { $set: { "grades.$.std" : 6 } } )
Note: Do not use the positional operator $ (page 788) with upsert operations because inserts will use the $ as a eld name in the inserted document. When used with the $unset (page 788) operator, the positional $ (page 788) operator does not remove the matching element from the array but rather sets it to null. See also: update() (page 948), $set (page 788) and $unset (page 788) $addToSet
789
$addToSet The $addToSet (page 789) operator adds a value to an array only if the value is not in the array already. If the value is in the array, $addToSet (page 789) returns without modifying the array. Consider the following example:
db.collection.update( { field: value }, { $addToSet: { field: value1 } } );
Here, $addToSet (page 789) appends value1 to the array stored in field, only if value1 is not already a member of this array. Note: $addToSet (page 789) only ensures that there are no duplicate items added to the set and does not affect existing duplicate elements. $addToSet (page 789) does not guarantee a particular ordering of elements in the modied set. :operator:$each Use the $each (page 793) modier with the $addToSet (page 789) operator to add multiple values to an array <field> if the values do not exist in the <field>.
db.collection.update( <query>, { $addToSet: { <field>: { $each: [ <value1>, <value2> ... ] } } } )
See also: $push (page 792) $pop $pop The $pop (page 790) operator removes the rst or last element of an array. Pass $pop (page 790) a value of 1 to remove the last element in an array and a value of -1 to remove the rst element of an array. Consider the following syntax:
db.collection.update( {field: value }, { $pop: { field: 1 } } );
This operation removes the last item of the array in field in the document that matches the query statement { field: value }. The following example removes the rst item of the same array:
db.collection.update( {field: value }, { $pop: { field: -1 } } );
Be aware of the following $pop (page 790) behaviors: The $pop (page 790) operation fails if field is not an array. $pop (page 790) will successfully remove the last item in an array. field will then hold an empty array. New in version 1.1. $pullAll $pullAll The $pullAll (page 790) operator removes multiple values from an existing array. $pullAll (page 790) provides the inverse operation of the $pushAll (page 791) operator. Consider the following example:
db.collection.update( { field: value }, { $pullAll: { field1: [ value1, value2, value3 ] } } );
790
Here, $pullAll (page 790) removes [ value1, value2, value3 ] from the array in field1, in the document that matches the query statement { field: value } in collection. $pull $pull The $pull (page 791) operator removes all instances of a value from an existing array, as in the following prototype:
db.collection.update( { field: <query> }, { $pull: { field: <query> } } );
$pull (page 791) removes items from the array in the eld named field that match the query in the $pull (page 791) statement. If a value (i.e. <value>) exists multiple times in an array, $pull (page 791) will remove all instances of the value. Example Given the following document in the cpuinfo collection:
{ flags: [vme, de, pse, tsc, msr, pae, mce ] }
The following operation will remove the msr value from the flags array:
db.cpuinfo.update( { flags: msr }, { $pull: { flags: msr } } )
Example Given the following document in the profiles collection:

{ votes: [ 3, 5, 6, 7, 7, 8 ] }
The following operation will remove all occurrences of 7 from the votes array.
db.profiles.update( { votes: 3 }, { $pull: { votes: 7 } } )
Therefore, the votes array would resemble the following: { votes: [ 3, 5, 6, 8 ] } Conversely, the following operation will remove all items from the array that are larger than 6:
db.profiles.update( { votes: 3 }, { $pull: { votes: { $gt: 6 } } } )
Therefore, the votes array would resemble the following:

{ votes: [ 3, 5, 6 ] }
$pushAll $pushAll Deprecated since version 2.4: Use the $push (page 792) operator with $each (page 793) instead. The $pushAll (page 791) operator is similar to the $push (page 792) but adds the ability to append several values to an array at once.
db.collection.update( { field: value }, { $pushAll: { field1: [ value1, value2, value3 ] } } );
791
Here, $pushAll (page 791) appends the values in [ value1, value2, value3 ] to the array in field1 in the document matched by the statement { field: value } in collection. If you specify a single value, $pushAll (page 791) will behave as $push (page 792). $push $push The $push (page 792) operator appends a specied value to an array.
db.collection.update( <query>, { $push: { <field>: <value> } } )
The following example appends 89 to the scores array for the rst document where the name eld equals joe:
db.students.update( { name: "joe" }, { $push: { scores: 89 } } )
Note: If the eld is absent in the document to update, $push (page 792) adds the array eld with the value as its element. If the eld is not an array, the operation will fail. If the value is an array, $push (page 792) appends the whole array as a single element. To add each element of the value separately, use $push (page 792) with the $each (page 793) modier. The following example appends each element of [ 90, 92, 85 ] to the scores array for the document where the name eld equals joe:
db.students.update( { name: "joe" }, { $push: { scores: { $each: [ 90, 92, 85 ] } } } )
Changed in version 2.4: MongoDB adds support for the $each (page 793) modier to the $push (page 792) operator. Before 2.4, use $pushAll (page 791) for similar functionality. Changed in version 2.4: You can use the $push (page 792) operator with the following modiers: $each (page 793) appends multiple values to the array eld, $slice (page 793), which is only available with $each (page 793), limits the number of array elements, and $sort (page 794), which is only available with $each (page 793), orders elements of the array. $sort (page 794) can only order array elements that are documents. The following example uses: the $each (page 793) modier to append documents to the quizzes array, the $sort (page 794) modier to sort all the elements of the modied quizzes array by the ascending score eld, and the $slice (page 793) modier to keep only the last ve sorted elements of the quizzes array.
792
db.students.update( { name: "joe" }, { $push: { quizzes: { $each: [ { id: 3, { id: 4, { id: 5, $sort: { score: 1 $slice: -5 } } } )
score: 8 }, score: 7 }, score: 6 } ], },
Update Operator Modiers
Name $each (page 793) $slice (page 793) $sort (page 794)
Description Modies the $push (page 792) and $addToSet (page 789) operators to append items for array updates. Modies the $push (page 792) operator to limit the size of updated arrays. Modies the $push (page 792) operator to reorder documents stored in an array.
$each Note: The $each (page 793) modier is only used with the $addToSet (page 789) and $push (page 792) operators. See the documentation of $addToSet (page 789) and $push (page 792) for more information. $each The $each (page 793) modier is available for use with the $addToSet (page 789) operator and the $push (page 792) operator. Use the $each (page 793) modier with the $addToSet (page 789) operator to add multiple values to an array <field> if the values do not exist in the <field>.
db.collection.update( <query>, { $addToSet: { <field>: { $each: [ <value1>, <value2> ... ] } } } )
Use the $each (page 793) modier with the $push (page 792) operator to append multiple values to an array <field>.
db.collection.update( <query>, { $push: { <field>: { $each: [ <value1>, <value2> ... ] } } } )
Changed in version 2.4: MongoDB adds support for the $each (page 793) modier to the $push (page 792) operator. $slice $slice New in version 2.4. The $slice (page 793) modier limits the number of array elements during a $push (page 792) operation. To project, or return, a specied number of array elements from a read operation, see the $slice (page 801) projection operator instead.
793
To use the $slice (page 793) modier, it must appear with the $each (page 793) modier, and the $each (page 793) modier must be the rst modier for the $push (page 792) operation.
db.collection.update( <query>, { $push: { <field>: { $each: [ <value1>, <value2>, ... ], $slice: <num> } } } )
The <num> is either a negative number or zero: If <num> is negative, the array <field> contains only the last <num> elements. If <num> is zero, the array <field> is an empty array.
db.students.update( { _id: 2 }, { $push: { grades: { $each: [ 80, 78, 86 ], $slice: -5 } } } )
$sort $sort New in version 2.4. The $sort (page 794) modier orders the elements of an array during a $push (page 792) operation. The elements of the array must be documents. $sort (page 794) modies $push (page 792) updates that use the $each (page 793) and $slice (page 793) modiers, where $each (page 793) is the rst modier for the $push (page 792) operation.
db.collection.update( <query>, { $push: { <field>: { $each: [ <document1>, <document2>, ... ], $slice: <num>, $sort: <sort document>, } } } )
Important: The <sort document> only accesses the elds from the elements in the array and does not refer to the array <field>. Consider the following example where the collection students contain the following document:
794
{ "_id": 3, "name": "joe", "quizzes": [ { "id" : 1, "score" : 6 }, { "id" : 2, "score" : 9 } ] }
The following update appends additional documents to the quizzes array, sorts all the elements of the array by ascending score eld, and slices the array to keep the last ve elements:
db.students.update( { name: "joe" }, { $push: { quizzes: { $each: [ { id: 3, { id: 4, { id: 5, $sort: { score: 1 $slice: -5 } } } ) score: 8 }, score: 7 }, score: 6 } ], },
After the update, the array elements are in order of ascending score eld.:
{ "_id" : 3, "name" : "joe", "quizzes" : [ { { { { { ] }
"id" "id" "id" "id" "id"
: : : : :
1, 5, 4, 3, 2,
"score" "score" "score" "score" "score"
: : : : :
6 6 7 8 9
}, }, }, }, }
Bitwise
Bitwise Update Operator
Name $bit (page 795) $bit $bit
Description Performs bitwise AND and OR updates of integer values.
The $bit (page 795) operator performs a bitwise update of a eld. Only use this with integer elds, as in the following examples:
db.collection.update( { field: NumberInt(1) }, { $bit: { field: { and: NumberInt(5) } } } ); db.collection.update( { field: NumberInt(1) }, { $bit: { field: { or: NumberInt(5) } } } );
Here, the $bit (page 795) operator updates the integer value of the eld named field: in the rst example with a bitwise and: 5 operation; and in the second example with a bitwise or: 5 operation. $bit (page 795) only works with integers.
795
$bit (page 795) only supports AND and OR bitwise operations. Note: All numbers in the mongo (page 1036) shell are doubles, not integers. Use the NumberInt() constructor to specify integers. See NumberInt (page 589) for more information.
Isolation
Isolation Update Operator
Name $isolated (page 796)
Description Modies behavior of multi-updates to improve the isolation of the operation.
$isolated $isolated $isolated (page 796) isolation operator isolates a write operation that affects multiple documents from other write operations. Note: The $isolated (page 796) isolation operator does not provide all-or-nothing atomicity for write operations. Consider the following example:
db.foo.update( { field1 : 1 , $isolated : 1 }, { $inc : { field2 : 1 } } , { multi: true } )
Without the $isolated (page 796) operator, multi-updates will allow other operations to interleave with this updates. If these interleaved operations contain writes, the update operation may produce unexpected results. By specifying $isolated (page 796) you can guarantee isolation for the entire multi-update. Warning: $isolated (page 796) does not work with sharded clusters. See also: See db.collection.update() (page 948) for more information db.collection.update() (page 948) method. $atomic Deprecated since version 2.2: The $isolated (page 796) replaces $atomic (page 796). about the
63.1.3 Projection Operators

Projection Operators Name $ (page 797) $elemMatch (page 798) $slice (page 801) Description Projects the rst element in an array that matches the query condition. Projects only the rst element from an array that matches the specied $elemMatch (page 798) condition. Limits the number of elements projected from an array. Supports skip and limit slices.
796
$ (projection)
$ The positional $ (page 797) operator limits the contents of the <array> eld that is included in the query results to contain the rst matching element. To specify an array element to update, see the positional $ operator for updates (page 788). Used in the projection document of the find() (page 924) method or the findOne() (page 929) method: The $ (page 797) projection operator limits the content of the <array> eld to the rst element that matches the query document (page 168). The <array> eld must appear in the query document (page 168)
db.collection.find( { { db.collection.find( { { <array>: <value> ... }, "<array>.$": 1 } ) <array.field>: <value> ...}, "<array>.$": 1 } )
The <value> can be documents that contains query operator expressions (page 763). Only one positional $ (page 797) operator can appear in the projection document. Only one array eld can appear in the query document (page 168); i.e. the following query is incorrect:
db.collection.find( { <array>: <value>, <someOtherArray>: <value2> }, { "<array>.$": 1 } )
Example A collection students contains the following documents:

{ { { { { { "_id" "_id" "_id" "_id" "_id" "_id" : : : : : : 1, 2, 3, 4, 5, 6, "semester" "semester" "semester" "semester" "semester" "semester" : : : : : : 1, 1, 1, 2, 2, 2, "grades" "grades" "grades" "grades" "grades" "grades" : : : : : : [ [ [ [ [ [ 70, 90, 85, 79, 88, 95, 87, 90 ] } 88, 92 ] } 100, 90 ] } 85, 80 ] } 88, 92 ] } 90, 96 ] }
In the following query, the projection { "grades.$": equal to 85 for the grades eld.
1 } returns only the rst element greater than or
db.students.find( { semester: 1, grades: { $gte: 85 } }, { "grades.$": 1 } )
The operation returns the following documents:

{ "_id" : 1, "grades" : [ 87 ] } { "_id" : 2, "grades" : [ 90 ] } { "_id" : 3, "grades" : [ 85 ] }
Although the array eld grades may contain multiple elements that are greater than or equal to 85, the $ (page 797) projection operator returns only the rst matching element from the array. Important: When the find() (page 924) method includes a sort() (page 965), the find() (page 924) method applies the sort() (page 965) to order the matching documents before it applies the positional $ (page 797) projection operator.
797
If an array eld contains multiple documents with the same eld name and the find() (page 924) method includes a sort() (page 965) on that repeating eld, the returned documents may not reect the sort order because the sort was applied to the elements of the array before the $ (page 797) projection operator. Example A students collection contains the following documents where the grades eld is an array of documents; each document contain the three eld names grade, mean, and std:
{ "_id" : 7, semester: 3, "grades" : [ { grade: 80, mean: 75, std: 8 }, { grade: 85, mean: 90, std: 5 }, { grade: 90, mean: 85, std: 3 } ] } { "_id" : 8, semester: 3, "grades" : [ { grade: 92, mean: 88, std: 8 }, { grade: 78, mean: 90, std: 5 }, { grade: 88, mean: 85, std: 3 } ] }
In the following query, the projection { "grades.$": 1 } returns only the rst element with the mean greater than 70 for the grades eld. The query also includes a sort() (page 965) to order by ascending grades.grade eld:
db.students.find( { "grades.mean": { $gt: 70 } }, { "grades.$": 1 } ).sort( { "grades.grade": 1 } )
The find() (page 924) method sorts the matching documents before it applies the $ (page 797) projection operator on the grades array. Thus, the results with the projected array elements do not reect the ascending grades.grade sort order:
{ "_id" : 8, "grades" : [ { "grade" : 92, "mean" : 88, "std" : 8 } ] } { "_id" : 7, "grades" : [ { "grade" : 80, "mean" : 75, "std" : 8 } ] }
Note: Since only one array eld can appear in the query document, if the array contains documents, to specify criteria on multiple elds of these documents, use the $elemMatch (query) (page 783) operator, e.g.:
db.students.find( { grades: { $elemMatch: { mean: { $gt: 70 }, grade: { $gt:90 } } } }, { "grades.$": 1 } )
See also: $elemMatch (projection) (page 798)

$elemMatch (projection)
See also: $elemMatch (query) (page 783) $elemMatch New in version 2.2. The $elemMatch (page 798) projection operator limits the contents of an array eld that is included in the query results to contain only the array element that matches the $elemMatch (page 798) condition.
798
Note: The elements of the array are documents. If multiple elements match the $elemMatch (page 798) condition, the operator returns the rst matching element in the array. The $elemMatch (page 798) projection operator is similar to the positional $ (page 797) projection operator. The examples on the $elemMatch (page 798) projection operator assumes a collection school with the following documents:
{ _id: 1, zipcode: 63109, students: [ { name: { name: { name: ] } { _id: 2, zipcode: 63110, students: [ { name: { name: ] }
"john", school: 102, age: 10 }, "jess", school: 102, age: 11 }, "jeff", school: 108, age: 15 }
"ajax", school: 100, age: 7 }, "achilles", school: 100, age: 8 },
{ _id: 3, zipcode: 63109, students: [ { name: "ajax", school: 100, age: 7 }, { name: "achilles", school: 100, age: 8 }, ] } { _id: 4, zipcode: 63109, students: [ { name: "barney", school: 102, age: 7 }, ] }
Example The following find() (page 924) operation queries for all documents where the value of the zipcode eld is 63109. The $elemMatch (page 798) projection returns only the rst matching element of the students array where the school eld has a value of 102:
db.schools.find( { zipcode: 63109 }, { students: { $elemMatch: { school: 102 } } } )
The operation returns the following documents:
799
{ "_id" : 1, "students" : [ { "name" : "john", "school" : 102, "age" : 10 } ] } { "_id" : 3 } { "_id" : 4, "students" : [ { "name" : "barney", "school" : 102, "age" : 7 } ] }
For the document with _id equal to 1, the students array contains multiple elements with the school eld equal to 102. However, the $elemMatch (page 798) projection returns only the rst matching element from the array. The document with _id equal to 3 does not contain the students eld in the result since no element in its students array matched the $elemMatch (page 798) condition. The $elemMatch (page 798) projection can specify criteria on multiple elds: Example The following find() (page 924) operation queries for all documents where the value of the zipcode eld is 63109. The projection includes the rst matching element of the students array where the school eld has a value of 102 and the age eld is greater than 10:
db.schools.find( { zipcode: 63109 }, { students: { $elemMatch: { school: 102, age: { $gt: 10} } } } )
The operation returns the three documents that have zipcode equal to 63109:
{ "_id" : 1, "students" : [ { "name" : "jess", "school" : 102, "age" : 11 } ] } { "_id" : 3 } { "_id" : 4 }
Documents with _id equal to 3 and _id equal to 4 do not contain the students eld since no element matched the $elemMatch (page 798) criteria. When the find() (page 924) method includes a sort() (page 965), the find() (page 924) method applies the sort() (page 965) to order the matching documents before it applies the projection. If an array eld contains multiple documents with the same eld name and the find() (page 924) method includes a sort() (page 965) on that repeating eld, the returned documents may not reect the sort order because the sort() (page 965) was applied to the elements of the array before the $elemMatch (page 798) projection. Example The following query includes a sort() (page 965) to order by descending students.age eld:
db.schools.find( { zipcode: 63109 }, { students: { $elemMatch: { school: 102 } } } ).sort( { "students.age": -1 } )
The operation applies the sort() (page 965) to order the documents that have the eld zipcode equal to 63109 and then applies the projection. The operation returns the three documents in the following order:
{ "_id" : 1, "students" : [ { "name" : "john", "school" : 102, "age" : 10 } ] } { "_id" : 3 } { "_id" : 4, "students" : [ { "name" : "barney", "school" : 102, "age" : 7 } ] }
See also: $ (projection) (page 797) operator 800 Chapter 63. MongoDB Interface
$slice (projection)
$slice The $slice (page 801) operator controls the number of items of an array that a query returns. For information on limiting the size of an array during an update with $push (page 792), see the $slice (page 793) modier instead. Consider the following prototype query:
db.collection.find( { field: value }, { array: {$slice: count } } );
This operation selects the document collection identied by a eld named field that holds value and returns the number of elements specied by the value of count from the array stored in the array eld. If count has a value greater than the number of elements in array the query returns all elements of the array. $slice (page 801) accepts arguments in a number of formats, including negative values and arrays. Consider the following examples:
db.posts.find( {}, { comments: { $slice: 5 } } )
Here, $slice (page 801) selects the rst ve items in an array in the comments eld.
db.posts.find( {}, { comments: { $slice: -5 } } )
This operation returns the last ve items in array. The following examples specify an array as an argument to $slice (page 801). Arrays take the form of [ skip , limit ], where the rst value indicates the number of items in the array to skip and the second value indicates the number of items to return.
db.posts.find( {}, { comments: { $slice: [ 20, 10 ] } } )
Here, the query will only return 10 items, after skipping the rst 20 items of that array.
db.posts.find( {}, { comments: { $slice: [ -20, 10 ] } } )
This operation returns 10 items as well, beginning with the item that is 20th from the last item of the array.
63.1.4 Projection Operators

Meta Query Operator Quick Reference
Introduction
In addition to the MongoDB Query Operators (page 763), there are a number of meta operators that you can modify the output or behavior of a query. On the server, MongoDB treats the query and the options as a single object. The mongo (page 1036) shell and driver interfaces may provide cursor methods (page 951) that wrap these options. When possible, use these methods; otherwise, you can add these options using either of the following syntax:
db.collection.find( { <query> } )._addSpecial( <option> ) db.collection.find( { $query: { <query> }, <option> } )
Operators
Modiers Many of these operators have corresponding methods in the shell (page 951). These methods provide a straightforward and user-friendly interface and are the preferred way to add these options. 63.1. Query, Update and Projection Operators 801
Name $comment (page 802) $explain (page 802) $hint (page 803) $maxScan (page 803) $max (page 804) $min (page 804) $orderby (page 805) $returnKey (page 806) $showDiskLoc (page 806) $snapshot (page 806)
Description Adds a comment to the query to identify queries in the database proler output. Forces MongoDB to report on query execution plans. See explain() (page 953). Forces MongoDB to use a specic index. See hint() (page 958) Limits the number of documents a cursor will return for a query. See limit() (page 959). Species a minimum exclusive upper limit for the index to use in a query. See max() (page 959). Species a minimum inclusive lower limit for the index to use in a query. See min() (page 961). Returns a cursor with documents sorted according to a sort specication. See sort() (page 965). Forces the cursor to only return elds included in the index. Modies the documents returned to include references to the on- disk location of each document. Forces the query to use the index on the _id eld. See snapshot() (page 964).
$comment $comment The $comment (page 802) makes it possible to attach a comment to a query. Because these comments propagate to the profile (page 888) log, adding $comment (page 802) modiers can make your prole data much easier to interpret and trace. Use one of the following forms:
db.collection.find( { <query> } )._addSpecial( "$comment", <comment> ) db.collection.find( { $query: { <query> }, $comment: <comment> } )
$explain $explain The $explain (page 802) operator provides information on the query plan. It returns a document that describes the process and indexes used to return the query. This may provide useful insight when attempting to optimize a query. For details on the output, see cursor.explain() (page 953). You can specify the $explain (page 802) operator in either of the following forms:
db.collection.find()._addSpecial( "$explain", 1 ) db.collection.find( { $query: {}, $explain: 1 } )
You also can specify $explain (page 802) through the explain() (page 953) method in the mongo (page 1036) shell:
db.collection.find().explain()
$explain (page 802) runs the actual query to determine the result. Although there are some differences between running the query with $explain (page 802) and running without, generally, the performance will be similar between the two. So, if the query is slow, the $explain (page 802) operation is also slow. Additionally, the $explain (page 802) operation reevaluates a set of candidate query plans, which may cause the $explain (page 802) operation to perform differently than a normal query. As a result, these operations generally provide an accurate account of how MongoDB would perform the query, but do not reect the length of these queries. To determine the performance of a particular index, you can use hint() (page 958) and in conjunction with explain() (page 953), as in the following example:
802
db.products.find().hint( { type: 1 } ).explain()
When you run explain() (page 953) with hint() (page 958), the query optimizer does not reevaluate the query plans. Note: In some situations, the explain() (page 953) operation may differ from the actual query plan used by MongoDB in a normal query. The explain() (page 953) operation evaluates the set of query plans and reports on the winning plan for the query. In normal operations the query optimizer caches winning query plans and uses them for similar related queries in the future. As a result MongoDB may sometimes select query plans from the cache that are different from the plan displayed using explain() (page 953). See also: cursor.explain() (page 953) Optimization Strategies for MongoDB (page 555) page for information regarding optimization strategies. Analyze Performance of Database Operations (page 99) tutorial for information regarding the database prole. Current Operation Reporting (page 971) $hint $hint The $hint (page 803) operator forces the query optimizer (page 174) to use a specic index to fulll the query. Specify the index either by the index name or by the index specication document. See Index Specication Documents (page 192) for information on index specication documents. Use $hint (page 803) for testing query performance and indexing strategies. The mongo (page 1036) shell provides a helper method hint() (page 958) for the $hint (page 803) operator. Consider the following operation:
db.users.find().hint( { age: 1 } )
This operation returns all documents in the collection named users using the index on the age eld. You can also specify a hint using either of the following forms:
db.users.find()._addSpecial( "$hint", { age : 1 } ) db.users.find( { $query: {}, $hint: { age : 1 } } )
Note: To combine $explain (page 802) and $hint (page 803) operations, use the following form:
db.users.find( { $query: {}, $hint: { age : 1 } } )
You must add the $explain (page 802) option to the document, as in the following:
db.users.find( { $query: {}, $hint: { age : 1 }, $explain: 1 } )
$maxScan $maxScan Constrains the query to only scan the specied number of documents when fullling the query. Use one of the following forms:
803
db.collection.find( { <query> } )._addSpecial( "$maxScan" , <number> ) db.collection.find( { $query: { <query> }, $maxScan: <number> } )
Use this modier to prevent potentially long running queries from disrupting performance by scanning through too much data. $max $max Specify a $max (page 804) value to specify the exclusive upper bound for a specic index in order to constrain the results of find() (page 924). The mongo (page 1036) shell provides the cursor.max() (page 959) wrapper method:
db.collection.find( { <query> } ).max( { field1: <max value>, ... fieldN: <max valueN> } )
You can also specify the option with either of the two forms:
db.collection.find( { <query> } )._addSpecial( "$max", { field1: <max value1>, ... fieldN: <max db.collection.find( { $query: { <query> }, $max: { field1: <max value1>, ... fieldN: <max valueN
The $max (page 804) species the upper bound for all keys of a specic index in order. Consider the following operations on a collection named collection that has an index { age:
db.collection.find( { <query> } ).max( { age: 100 } )
1 }:
This operation limits the query to those documents where the eld age is less than 100 using the index { age: 1 }. You can explicitly specify the corresponding index with cursor.hint() (page 958). Otherwise, MongoDB selects the index using the elds in the indexBounds; however, if multiple indexes exist on same elds with different sort orders, the selection of the index may be ambiguous. Consider a collection named collection that has the following two indexes:
{ age: 1, type: -1 } { age: 1, type: 1 }
Without explicitly using cursor.hint() (page 958), MongoDB may select either index for the following operation:
db.collection.find().max( { age: 50, type: B } )
Use $max (page 804) alone or in conjunction with $min (page 804) to limit results to a specic range for the same index, as in the following example:
db.collection.find().min( { age: 20 } ).max( { age: 25 } )
Note: Because cursor.max() (page 959) requires an index on a eld, and forces the query to use this index, you may prefer the $lt (page 766) operator for the query if possible. Consider the following example:
db.collection.find( { _id: 7 } ).max( { age: 25 } )
The query uses the index on the age eld, even if the index on _id may be better.
$min
804
$min Specify a $min (page 804) value to specify the inclusive lower bound for a specic index in order to constrain the results of find() (page 924). The mongo (page 1036) shell provides the cursor.min() (page 961) wrapper method:
db.collection.find( { <query> } ).min( { field1: <min value>, ... fieldN: <min valueN>} )
You can also specify the option with either of the two forms:
db.collection.find( { <query> } )._addSpecial( "$min", { field1: <min value1>, ... fieldN: <min db.collection.find( { $query: { <query> }, $min: { field1: <min value1>, ... fieldN: <min valueN
The $min (page 804) species the lower bound for all keys of a specic index in order. Consider the following operations on a collection named collection that has an index { age:
db.collection.find().min( { age: 20 } )
1 }:
These operations limit the query to those documents where the eld age is at least 20 using the index { age: 1 }. You can explicitly specify the corresponding index with cursor.hint() (page 958). Otherwise, MongoDB selects the index using the elds in the indexBounds; however, if multiple indexes exist on same elds with different sort orders, the selection of the index may be ambiguous. Consider a collection named collection that has the following two indexes:
{ age: 1, type: -1 } { age: 1, type: 1 }
Without explicitly using cursor.hint() (page 958), it is unclear which index the following operation will select:
db.collection.find().min( { age: 20, type: C } )
You can use $min (page 804) in conjunction with $max (page 804) to limit results to a specic range for the same index, as in the following example:
db.collection.find().min( { age: 20 } ).max( { age: 25 } )
Note: Because cursor.min() (page 961) requires an index on a eld, and forces the query to use this index, you may prefer the $gte (page 765) operator for the query if possible. Consider the following example:
db.collection.find( { _id: 7 } ).min( { age: 25 } )
The query will use the index on the age eld, even if the index on _id may be better.
$orderby $orderby The $orderby (page 805) operator sorts the results of a query in ascending or descending order. The mongo (page 1036) shell provides the cursor.sort() (page 965) method:
db.collection.find().sort( { age: -1 } )
You can also specify the option in either of the following forms:
db.collection.find()._addSpecial( "$orderby", { age : -1 } ) db.collection.find( { $query: {}, $orderby: { age : -1 } } )
805
These examples return all documents in the collection named collection sorted by the age eld in descending order. Specify a value to $orderby (page 805) of negative one (e.g. -1, as above) to sort in descending order or a positive value (e.g. 1) to sort in ascending order. Unless you have an index for the specied key pattern, use $orderby (page 805) in conjunction with $maxScan (page 803) and/or cursor.limit() (page 959) to avoid requiring MongoDB to perform a large in-memory sort. The cursor.limit() (page 959) increases the speed and reduces the amount of memory required to return this query by way of an optimized algorithm. $returnKey $returnKey Only return the index eld or elds for the results of the query. If $returnKey (page 806) is set to true and the query does not use an index to perform the read operation, the returned documents will not contain any elds. Use one of the following forms:
db.collection.find( { <query> } )._addSpecial( "$returnKey", true ) db.collection.find( { $query: { <query> }, $returnKey: true } )
$showDiskLoc $showDiskLoc $showDiskLoc (page 806) option adds a eld $diskLoc to the returned documents. The $diskLoc eld contains the disk location information. The mongo (page 1036) shell provides the cursor.showDiskLoc() (page 963) method:
db.collection.find().showDiskLoc()
db.collection.find( { <query> } )._addSpecial("$showDiskLoc" , true) db.collection.find( { $query: { <query> }, $showDiskLoc: true } )
$snapshot $snapshot The $snapshot (page 806) operator prevents the cursor from returning a document more than once because an intervening write operation results in a move of the document. Even in snapshot mode, objects inserted or deleted during the lifetime of the cursor may or may not be returned. The mongo (page 1036) shell provides the cursor.snapshot() (page 964) method:
db.collection.find().snapshot()
db.collection.find()._addSpecial( "$snapshot", true ) db.collection.find( { $query: {}, $snapshot: true } )
The $snapshot (page 806) operator traverses the index on the _id eld 1 . Warning: You cannot use $snapshot (page 806) with sharded collections. Do not use $snapshot (page 806) with $hint (page 803) or $orderby (page 805) (or the corresponding cursor.hint() (page 958) and cursor.sort() (page 965) methods.)
1
You can achieve the $snapshot (page 806) isolation behavior using any unique index on invariable elds.
806
Sort Order
Name $natural (page 807)
Description A special sort order that orders documents using the order of documents on disk.
$natural $natural Use the $natural (page 807) operator to use natural order for the results of a sort operation. Natural order refers to the order of documents in the le on disk. The $natural (page 807) operator uses the following syntax to return documents in the order they exist on disk:
db.collection.sort( { $natural: 1 } )
Use -1 to return documents in the reverse order as they occur on disk:

db.collection.sort( { $natural: -1 } )
See also: cursor.sort() (page 965)
63.2 Database Commands

User Commands (page 807) Aggregation Commands (page 807) Geospatial Commands (page 822) Query and Write Operation Commands (page 823) Database Operations (page 837) Replication Commands (page 837) Sharding Commands (page 846) Instance Administration Commands (page 855) Diagnostic Commands (page 871) Internal Commands (page 910) Testing Commands (page 913) All command documentation lined below describes the commands and available parameters, provides a document template or prototype for each command. Some command documentation also includes the relevant mongo (page 1036).
63.2.1 User Commands

Aggregation Commands
Aggregation Commands
Name aggregate (page 808) count (page 808) distinct (page 809) group (page 810) mapReduce (page 814)
Description Performs aggregation tasks (page 255) such as group using the aggregation framework. Counts the number of documents in a collection. Displays the distinct values found for a specied key in a collection. Groups documents in a collection by the specied key and performs simple aggregation. Performs map-reduce (page 311) aggregation for large data sets.
63.2. Database Commands
807
aggregate aggregate New in version 2.1.0. aggregate (page 808) implements the aggregation framework. Consider the following prototype form:
{ aggregate: "[collection]", pipeline: [pipeline] }
Where [collection] species the name of the collection that contains the data that you wish to aggregate. The pipeline argument holds an array that contains the specication for the aggregation operation. Consider the following example from the aggregation documentation (page 255).
db.runCommand( { aggregate : "article", pipeline : [ { $project : { author : 1, tags : 1, } }, { $unwind : "$tags" }, { $group : { _id : "$tags", authors : { $addToSet : "$author" } } } ] } );
More typically this operation would use the aggregate() (page 918) helper in the mongo (page 1036) shell, and would resemble the following:
db.article.aggregate( { $project : { author : 1, tags : 1, } }, { $unwind : "$tags" }, { $group : { _id : "$tags", authors : { $addToSet : "$author" } } } );
Changed in version 2.4: If an error occurs, the aggregate() (page 918) helper throws an exception. In previous versions, the helper returned a document with the error message and code, and ok status eld not equal to 1, same as the aggregate (page 808) command. For more aggregation documentation, please see: Aggregation Framework (page 255) Aggregation Framework Reference (page 271) Aggregation Framework Examples (page 261) count count The count (page 808) command counts the number of documents in a collection. The command returns a document that contains the count as well as the command status. The count (page 808) command takes the following prototype form:
808
{ count: <collection>, query: <query>, limit: <limit>, skip: <skip> }
The command elds are as follows: Fields count (String) The name of the collection to count. query (document) Optional. Species the selection query to determine which documents in the collection to count. limit (integer) Optional. Species the limit for the documents matching the selection query. skip (integer) Optional. Species the number of matching documents to skip. Consider the following examples of the count (page 808) command: Count the number of all documents in the orders collection:
db.runCommand( { count: orders } )
In the result, the n, which represents the count, is 26 and the command status ok is 1:
{ "n" : 26, "ok" : 1 }
Count the number of the documents in the orders collection with the eld ord_dt greater than new Date(01/01/2012):
db.runCommand( { count:orders, query: { ord_dt: { $gt: new Date(01/01/2012) } } } )
{ "n" : 13, "ok" : 1 }
Count the number of the documents in the orders collection with the eld ord_dt greater than new Date(01/01/2012) skipping the rst 10 matching records:
db.runCommand( { count:orders, query: { ord_dt: { $gt: new Date(01/01/2012) } }, skip: 10 } )
{ "n" : 3, "ok" : 1 }
Note: MongoDB also provides the cursor.count() (page 952) method and the shell wrapper db.collection.count() (page 919) method.
distinct distinct The distinct (page 809) command nds the distinct values for a specied eld across a single collection. The command returns a document that contains an array of the distinct values as well as the query plan and status. The command takes the following prototype form:
{ distinct: collection, key: <field>, query: <query> }
The command elds are as follows:
809
Fields collection (String) The name of the collection to query for distinct values. eld (string) Species the eld for which to return the distinct values. query (document) Optional. Species the selection query to determine the subset of documents from which to retrieve the distinct values. Consider the following examples of the distinct (page 809) command: Return an array of the distinct values of the eld ord_dt from all documents in the orders collection:
db.runCommand ( { distinct: orders, key: ord_dt } )
Return an array of the distinct values of the eld sku in the subdocument item from all documents in the orders collection:
db.runCommand ( { distinct: orders, key: item.sku } )
Return an array of the distinct values of the eld ord_dt from the documents in the orders collection where the price is greater than 10:
db.runCommand ( { distinct: orders, key: ord_dt, query: { price: { $gt: 10 } } } )
Note: MongoDB also provides the shell wrapper method db.collection.distinct() (page 920) for the distinct (page 809) command. Additionally, many MongoDB drivers also provide a wrapper method. Refer to the specic driver documentation. When possible, the distinct (page 809) command will use an index to nd the documents in the query as well as to return the data.
group group The group (page 810) command groups documents in a collection by the specied key and performs simple aggregation functions such as computing counts and sums. The command is analogous to a SELECT ... GROUP BY statement in SQL. The command returns a document with the grouped records as well as the command meta-data. The group (page 810) command takes the following prototype form:
{ group: { ns: <namespace>, key: <key>, $reduce: <reduce function>, $keyf: <key function>, cond: <query>, finalize: <finalize function> } }
The command elds are as follows: Fields ns Species the collection from which to perform the group by operation. key Species one or more document elds to group. Returns a key object for use as the grouping key. 810 Chapter 63. MongoDB Interface
$reduce Species an aggregation function that operates on the documents during the grouping operation, such as compute a sum or a count. The aggregation function takes two arguments: the current document and an aggregation result document for that group. initial Initializes the aggregation result document. $keyf Optional. Alternative to the key eld. Species a function that creates a key object for use as the grouping key. Use the keyf instead of key to group by calculated elds rather than existing document elds. cond Optional. Species the selection criteria to determine which documents in the collection to process. If the cond eld is omitted, the db.collection.group() (page 932) processes all the documents in the collection for the group operation. nalize Optional. Species a function that runs each item in the result set before db.collection.group() (page 932) returns the nal value. This function can either modify the result document or replace the result document as a whole. Note: Unlike the $keyf and the $reduce elds that specify a function, the eld name is finalize and not $finalize. Warning: The group (page 810) command does not work with sharded clusters. Use the aggregation framework or map-reduce in sharded environments. The result set must t within the maximum BSON document size (page 1105). Additionally, in version 2.2, the returned array can contain at most 20,000 elements; i.e. at most 20,000 unique groupings. For group by operations that results in more than 20,000 unique groupings, use mapReduce (page 814). Previous versions had a limit of 10,000 elements. Prior to 2.4, the group (page 810) command took the mongod (page 1021) instances JavaScript lock which blocked all other JavaScript execution.
Note: Changed in version 2.4. In MongoDB 2.4, map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions cannot access certain global functions or properties, such as db, that are available in the mongo (page 1036) shell. When upgrading to MongoDB 2.4, you will need to refactor your code if your map-reduce operations (page 814), group (page 810) commands, or $where (page 775) operator expressions include any global shell functions or properties that are no longer available, such as db. The following shell functions and properties are available to map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions in MongoDB 2.4:
811
For the shell, MongoDB provides a wrapper method db.collection.group() (page 932); however, the db.collection.group() (page 932) method takes the keyf eld and the reduce eld whereas the group (page 810) command takes the $keyf eld and the $reduce eld. JavaScript in MongoDB Although group (page 810) uses JavaScript, most interactions with MongoDB do not use JavaScript but use an idiomatic driver (page 555) in the language of the interacting application. The following are examples of the db.collection.group() (page 932) method: The examples assume an orders collection with documents of the following prototype:
{ _id: ObjectId("5085a95c8fada716c89d0021"), ord_dt: ISODate("2012-07-01T04:00:00Z"), ship_dt: ISODate("2012-07-02T04:00:00Z"), item: { sku: "abc123", price: 1.99, uom: "pcs", qty: 25 } }
The following example groups by the ord_dt and item.sku elds those documents that have ord_dt greater than 01/01/2012:
db.runCommand( { group: { ns: orders, key: { ord_dt: 1, item.sku: 1 }, cond: { ord_dt: { $gt: new Date( 01/01/2012 ) } }, $reduce: function ( curr, result ) { }, initial: { } } } )
812
The result is a documents that contain the retval eld which contains the group by records, the count eld which contains the total number of documents grouped, the keys eld which contains the number of unique groupings (i.e. number of elements in the retval), and the ok eld which contains the command status:
{ "retval" : [ { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" ], "count" : 13, "keys" : 11, "ok" : 1 } : : : : : : : : : : : ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-05-01T04:00:00Z"), ISODate("2012-05-01T04:00:00Z"), ISODate("2012-06-08T04:00:00Z"), ISODate("2012-06-08T04:00:00Z"), "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" : : : : : : : : : : : "abc123"}, "abc456"}, "bcd123"}, "efg456"}, "abc123"}, "efg456"}, "ijk123"}, "abc123"}, "abc456"}, "abc123"}, "abc456"}
The method call is analogous to the SQL statement:

SELECT ord_dt, item_sku FROM orders WHERE ord_dt > 01/01/2012 GROUP BY ord_dt, item_sku
The following example groups by the ord_dt and item.sku elds, those documents that have ord_dt greater than 01/01/2012 and calculates the sum of the qty eld for each grouping:
db.runCommand( { group: { ns: orders, key: { ord_dt: 1, item.sku: 1 }, cond: { ord_dt: { $gt: new Date( 01/01/2012 ) } }, $reduce: function ( curr, result ) { result.total += curr.item.qty; }, initial: { total : 0 } } } )
The retval eld of the returned document is an array of documents that contain the group by elds and the calculated aggregation eld:
{ "retval" : [ { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" { "ord_dt" : : : : : : : : : : : ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-05-01T04:00:00Z"), ISODate("2012-05-01T04:00:00Z"), ISODate("2012-06-08T04:00:00Z"), ISODate("2012-06-08T04:00:00Z"), "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" : : : : : : : : : : : "abc123", "abc456", "bcd123", "efg456", "abc123", "efg456", "ijk123", "abc123", "abc456", "abc123", "abc456", "total" "total" "total" "total" "total" "total" "total" "total" "total" "total" "total" : : : : : : : : : : : 25 25 10 10 25 15 20 45 25 25 25
}, }, }, }, }, }, }, }, }, }, }
813
], "count" : 13, "keys" : 11, "ok" : 1 }

SELECT ord_dt, item_sku, SUM(item_qty) as total FROM orders WHERE ord_dt > 01/01/2012 GROUP BY ord_dt, item_sku
The following example groups by the calculated day_of_week eld, those documents that have ord_dt greater than 01/01/2012 and calculates the sum, count, and average of the qty eld for each grouping:
db.runCommand( { group: { ns: orders, $keyf: function(doc) { return { day_of_week: doc.ord_dt.getDay() } ; }, cond: { ord_dt: { $gt: new Date( 01/01/2012 ) } }, $reduce: function ( curr, result ) { result.total += curr.item.qty; result.count++; }, initial: { total : 0, count: 0 }, finalize: function(result) { var weekdays = [ "Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday" ]; result.day_of_week = weekdays[result.day_of_week]; result.avg = Math.round(result.total / result.count); } } } )
The retval eld of the returned document is an array of documents that contain the group by elds and the calculated aggregation eld:
{ "retval" : [ { "day_of_week" : "Sunday", "total" : 70, "count" : 4, "avg" : 18 }, { "day_of_week" : "Friday", "total" : 110, "count" : 6, "avg" : 18 }, { "day_of_week" : "Tuesday", "total" : 70, "count" : 3, "avg" : 23 } ], "count" : 13, "keys" : 3, "ok" : 1 }
See also: Aggregation Framework (page 255) mapReduce
814
mapReduce The mapReduce (page 814) command allows you to run map-reduce aggregation operations over a collection. The mapReduce (page 814) command has the following prototype form:
db.runCommand( { mapReduce: <collection>, map: <function>, reduce: <function>, out: <output>, query: <document>, sort: <document>, limit: <number>, finalize: <function>, scope: <document>, jsMode: <boolean>, verbose: <boolean> } )
Pass the name of the collection to the mapReduce command (i.e. <collection>) to use as the source documents to perform the map reduce operation. The command also accepts the following parameters: Parameters map A JavaScript function that associates or maps a value with a key and emits the key and value pair. The map function processes every input document for the map-reduce operation. The mapreduce operation groups the emitted value objects by the key and passes these groupings to the reduce function. reduce A JavaScript function that reduces to a single object all the values associated with a particular key. The reduce function accepts two arguments: key and values. The values argument is an array whose elements are the value objects that are mapped to the key. out New in version 1.8. Species the location of the result of the map-reduce operation. You can output to a collection, output to a collection with an action, or output inline. You may output to a collection when performing map reduce operations on the primary members of the set; on secondary members you may only use the inline output. query Optional. Species the selection criteria using query operators (page 763) for determining the documents input to the map function. sort Optional. Sorts the input documents. This option is useful for optimization. For example, specify the sort key to be the same as the emit key so that there are fewer reduce operations. limit Optional. Species a maximum number of documents to return from the collection. nalize Optional. A JavaScript function that follows the reduce method and modies the output. The finalize function receives two arguments: key and reducedValue. reducedValue is the value returned from the reduce function for the key. The
scope (document) Optional. Species global variables that are accessible in the map , reduce and the finalize functions.
815
jsMode (Boolean) New in version 2.0. Optional. Species whether to convert intermediate data into BSON format between the execution of the map and reduce functions. If false: Internally, MongoDB converts the JavaScript objects emitted by the map function to BSON objects. These BSON objects are then converted back to JavaScript objects when calling the reduce function. The map-reduce operation places the intermediate BSON objects in temporary, on-disk storage. This allows the map-reduce operation to execute over arbitrarily large data sets. If true: Internally, the JavaScript objects emitted during map function remain as JavaScript objects. There is no need to convert the objects for the reduce function, which can result in faster execution. You can only use jsMode for result sets with fewer than 500,000 distinct key arguments to the mappers emit() function. The jsMode defaults to false. verbose (Boolean) Optional. Species whether to include the timing information in the result information. The verbose defaults to true to include the timing information. The following is a prototype usage of the mapReduce (page 814) command:
var mapFunction = function() { ... }; var reduceFunction = function(key, values) { ... }; db.runCommand( { mapReduce: orders, map: mapFunction, reduce: reduceFunction, out: { merge: map_reduce_results, db: test }, query: { ord_date: { $gt: new Date(01/01/2012) } } } )
JavaScript in MongoDB Although mapReduce (page 814) uses JavaScript, most interactions with MongoDB do not use JavaScript but use an idiomatic driver (page 555) in the language of the interacting application. Note: Changed in version 2.4. In MongoDB 2.4, map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions cannot access certain global functions or properties, such as db, that are available in the mongo (page 1036) shell. When upgrading to MongoDB 2.4, you will need to refactor your code if your map-reduce operations (page 814), group (page 810) commands, or $where (page 775) operator expressions include any global shell functions or properties that are no longer available, such as db. The following shell functions and properties are available to map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions in MongoDB 2.4:
816
Requirements for the map Function The map function has the following prototype:
function() { ... emit(key, value); }
The map function exhibits the following behaviors: In the map function, reference the current document as this within the function. The map function should not access the database for any reason. The map function should be pure, or have no impact outside of the function (i.e. side effects.) The emit(key,value) function associates the key with a value. A single emit can only hold half of MongoDBs maximum BSON document size (page 1105). There is no limit to the number of times you may call the emit function per document. The map function can access the variables dened in the scope parameter. Requirements for the reduce Function The reduce function has the following prototype:
function(key, values) { ... return result; }
The reduce function exhibits the following behaviors: The reduce function should not access the database, even to perform read operations. The reduce function should not affect the outside system. MongoDB will not call the reduce function for a key that has only a single value.
817
The reduce function can access the variables dened in the scope parameter. Because it is possible to invoke the reduce function more than once for the same key, the following properties need to be true: the type of the return object must be identical to the type of the value emitted by the map function to ensure that the following operations is true:
reduce(key, [ C, reduce(key, [ A, B ]) ] ) == reduce( key, [ C, A, B ] )
the reduce function must be idempotent. Ensure that the following statement is true:
reduce( key, [ reduce(key, valuesArray) ] ) == reduce( key, valuesArray )
the order of the elements in the valuesArray should not affect the output of the reduce function, so that the following statement is true:
reduce( key, [ A, B ] ) == reduce( key, [ B, A ] )
out Options You can specify the following options for the out parameter: Output to a Collection
out: <collectionName>
Output to a Collection with an Action This option is only available when passing out a collection that already exists. This option is not available on secondary members of replica sets.
out: { <action>: <collectionName> [, db: <dbName>] [, sharded: <boolean> ] [, nonAtomic: <boolean> ] }
When you output to a collection with an action, the out has the following parameters: <action>: Specify one of the following actions: replace Replace the contents of the <collectionName> if the collection with the <collectionName> exists. merge Merge the new result with the existing result if the output collection already exists. If an existing document has the same key as the new result, overwrite that existing document. reduce Merge the new result with the existing result if the output collection already exists. If an existing document has the same key as the new result, apply the reduce function to both the new and the existing documents and overwrite the existing document with the result. db: Optional.The name of the database that you want the map-reduce operation to write its output. By default this will be the same database as the input collection. sharded: Optional. If true and you have enabled sharding on output database, the map-reduce operation will shard the output collection using the _id eld as the shard key. 818 Chapter 63. MongoDB Interface
nonAtomic: New in version 2.2. Optional. Specify output operation as non-atomic and is valid only for merge and reduce output modes which may take minutes to execute. If nonAtomic is true, the post-processing step will prevent MongoDB from locking the database; however, other clients will be able to read intermediate states of the output collection. Otherwise the map reduce operation must lock the database during post-processing. Output Inline Perform the map-reduce operation in memory and return the result. This option is the only available option for out on secondary members of replica sets.
out: { inline: 1 }
The result must t within the maximum size of a BSON document (page 1105). Requirements for the finalize Function The finalize function has the following prototype:
function(key, reducedValue) { ... return modifiedObject; }
The finalize function receives as its arguments a key value and the reducedValue from the reduce function. Be aware that: The finalize function should not access the database for any reason. The finalize function should be pure, or have no impact outside of the function (i.e. side effects.) The finalize function can access the variables dened in the scope parameter. Examples In the mongo (page 1036) shell, the db.collection.mapReduce() (page 937) method is a wrapper around the mapReduce (page 814) command. The following examples use the db.collection.mapReduce() (page 937) method: Consider the following map-reduce operations on a collection orders that contains documents of the following prototype:
1. Dene the map function to process each input document: In the function, this refers to the document that the map-reduce operation is processing. 63.2. Database Commands 819
The function maps the price to the cust_id for each document and emits the cust_id and price pair.
This operation outputs the results to a collection named map_reduce_example. If the map_reduce_example collection already exists, the operation will replace the contents with the results of this map-reduce operation: Calculate Order and Total Quantity with Average Quantity Per Item In this example you will perform a map-reduce operation on the orders collection, for all documents that have an ord_date value greater than 01/01/2012. The operation groups by the item.sku eld, and for each sku calculates the number of orders and the total quantity ordered. The operation concludes by calculating the average quantity per order for each sku value: 1. Dene the map function to process each input document: In the function, this refers to the document that the map-reduce operation is processing. For each item, the function associates the sku with a new object value that contains the count of 1 and the item qty for the order and emits the sku and value pair.
2. Dene the corresponding reduce function with two arguments keySKU and countObjVals: countObjVals is an array whose elements are the objects mapped to the grouped keySKU values passed by map function to the reducer function.
820
The function reduces the countObjVals array to a single object reducedValue that contains the count and the qty elds. In reducedVal, the count eld contains the sum of the count elds from the individual array elements, and the qty eld contains the sum of the qty elds from the individual array elements.
using
the
mapFunction2,
This operation uses the query eld to select only those documents with ord_date greater than new Date(01/01/2012). Then it output the results to a collection map_reduce_example. If the map_reduce_example collection already exists, the operation will merge the existing contents with the results of this map-reduce operation: For more information and examples, see the Map-Reduce (page 311) page and Perform Incremental Map-Reduce (page 313). See also: Troubleshoot the Map Function (page 317) Troubleshoot the Reduce Function (page 318) db.collection.mapReduce() (page 937) Aggregation Framework (page 255)
821
Geospatial Commands
Geospatial Commands
Name geoNear (page 822) geoSearch (page 823) geoWalk (page 823)
Description Performs a geospatial query that returns the 100 documents closest to a given point. Performs a geospatial query that uses MongoDBs haystack index functionality. An internal command to support geospatial queries.
geoNear geoNear The geoNear (page 822) command species a point for which a geospatial query returns the 100 closest documents. The geoNear (page 822) command provides an alternative to the $near (page 780) operator. In addition to the functionality of $near (page 780), geoNear (page 822) returns additional diagnostic information. The geoNear (page 822) command can use either a GeoJSON point or legacy coordinate pairs. For a GeoJSON point, use the following syntax:
db.runCommand( { geoNear : <collection> , near : { type : "Point" , coordinates: [ <coordinates> ] } , spherical : true } )
For legacy coordinate pairs, use:

{ geoNear : <collection> , near : [ <coordinates> ] }
The geoNear (page 822) command provides the following options. Specify all distances in the same units as the document coordinate system: Fields near Can use either a GeoJSON point or legacy points, as shown above. limit Optional. Species the maximum number of documents to return. The default value is 100. See also the num option. num Optional. Synonym for the limit option. If both num and limit are included, the num value overrides the limit value. maxDistance Optional. Limits the results to those falling within a given distance of the center coordinate. For GeoJSON data distance is in meters. For grid coordinate (page 359) data distance is in radians. query Optional. Further narrows the results using any standard MongoDB query operator or selection. See db.collection.find() (page 924) and Query, Update and Projection Operators (page 763) for more information. spherical Optional. Default: false. When true MongoDB will return the query as if the coordinate system references points on a spherical plane rather than a plane. distanceMultiplier Optional. Species a factor to multiply all distances returned by geoNear (page 822). For example, use distanceMultiplier to convert from spherical queries returned in radians to linear units (i.e. miles or kilometers) by multiplying by the radius of the Earth. includeLocs Optional. Default: false. When specied true, the query will return the location of the matching documents in the result. 822 Chapter 63. MongoDB Interface
uniqueDocs Optional. Default true. The default settings will only return a matching document once, even if more than one of its location elds match the query. When false the query will return documents with multiple matching location elds more than once. See $uniqueDocs for more information on this option geoSearch geoSearch The geoSearch (page 823) command provides an interface to MongoDBs haystack index functionality. These indexes are useful for returning results based on location coordinates after collecting results based on some other query (i.e. a haystack.) Consider the following example:
{ geoSearch : "places", near : [33, 33], maxDistance : 6, search : { type : "restaurant" }, limi
The above command returns all documents with a type of restaurant having a maximum distance of 6 units from the coordinates [30,33] in the collection places up to a maximum of 30 results. Unless specied otherwise, the geoSearch (page 823) command limits results to 50 documents. geoWalk geoWalk geoWalk (page 823) is an internal command. Query and Write Operation Commands
Query and Write Operation Commands
Name eval (page 823) findAndModify (page 826) getLastError (page 831) getPrevError (page 832) resetError (page 832) text (page 832) eval eval
Description Runs a JavaScript function on the database server. Returns and modies a single document. Returns the success status of the last operation. Returns status document containing all errors since the last resetError (page 832) command. Resets the last error status. Performs a text search.
The eval (page 823) command evaluates JavaScript functions on the database server and has the following form:
{ eval: <function>, args: [ <arg1>, <arg2> ... ], nolock: <boolean> }
The command contains the following elds: Fields
823
eval (function) A JavaScript function. The function need not take any arguments, as in the rst example, or may take arguments as in the second:
function () { // ... } function (arg1, arg2) { // ... }
args (array) An array of corresponding arguments to pass to the specied JavaScript function if the function accepts arguments. Omit if the function does not take arguments. nolock (boolean) Optional. By default, eval (page 823) takes a global write lock before evaluating the JavaScript function. As a result, eval (page 823) blocks all other read and write operations to the database while the eval (page 823) operation runs. Set nolock to true on the eval (page 823) command to prevent the eval (page 823) command from taking the global write lock before evaluating the JavaScript. nolock does not impact whether operations within the JavaScript code itself takes a write lock. JavaScript in MongoDB Although eval (page 823) uses JavaScript, most interactions with MongoDB do not use JavaScript but use an idiomatic driver (page 555) in the language of the interacting application. The following example uses eval (page 823) to perform an increment and calculate the average on the server:
db.runCommand( { eval: function(name, incAmount) { var doc = db.myCollection.findOne( { name : name } ); doc = doc || { name : name , num : 0 , total : 0 , avg : 0 }; doc.num++; doc.total += incAmount; doc.avg = doc.total / doc.num; db.myCollection.save( doc ); return doc; }, args: [ "eliot", 5 ] } );
The db in the function refers to the current database. The mongo (page 1036) shell provides a helper method db.eval() (page 975) 2 , so you can express the above as follows:
2 The helper db.eval() (page 975) in the mongo (page 1036) shell wraps the eval (page 823) command. Therefore, the helper method shares the characteristics and behavior of the underlying command with one exception: db.eval() (page 975) method does not support the nolock option.
824
db.eval( function(name, incAmount) { var doc = db.myCollection.findOne( { name : name } ); doc = doc || { name : name , num : 0 , total : 0 , avg : 0 }; doc.num++; doc.total += incAmount; doc.avg = doc.total / doc.num; db.myCollection.save( doc ); return doc; }, "eliot", 5 );
If you want to use the servers interpreter, you must run eval (page 823). Otherwise, the mongo (page 1036) shells JavaScript interpreter evaluates functions entered directly into the shell. If an error occurs, eval (page 823) throws an exception. The following invalid function uses the variable x without declaring it as an argument:
db.runCommand( { eval: function() { return x + x; }, args: [ 3 ] } )
The statement will result in the following exception:

{ "errmsg" : "exception: JavaScript execution failed: ReferenceError: x is not defined near { "code" : 16722, "ok" : 0 }
Warning: By default, eval (page 823) takes a global write lock before evaluating the JavaScript function. As a result, eval (page 823) blocks all other read and write operations to the database while the eval (page 823) operation runs. Set nolock to true on the eval (page 823) command to prevent the eval (page 823) command from taking the global write lock before evaluating the JavaScript. nolock does not impact whether operations within the JavaScript code itself takes a write lock. Do not use eval (page 823) for long running operations as eval (page 823) blocks all other operations. Consider using other server side code execution options (page 560). You can not use eval (page 823) with sharded data. In general, you should avoid using eval (page 823) in sharded cluster; nevertheless, it is possible to use eval (page 823) with non-sharded collections and databases stored in a sharded cluster. With authentication (page 1081) enabled, eval (page 823) will fail during the operation if you do not have the permission to perform a specied task. Changed in version 2.4: You must have full admin access to run. Changed in version 2.4: The V8 JavaScript engine, which became the default in 2.4, allows multiple JavaScript operations to execute at the same time. Prior to 2.4, eval (page 823) executed in a single thread. See also: Server-side JavaScript (page 560)
825
ndAndModify findAndModify The findAndModify (page 826) command atomically modies and returns a single document. By default, the returned document does not include the modications made on the update. To return the document with the modications made on the update, use the new option. The command has the following syntax:
{ findAndModify: <string>, query: <document>, sort: <document>, remove: <boolean>, update: <document>, new: <boolean>, fields: <document>, upsert: <boolean> }
The findAndModify (page 826) command takes the following elds: Fields ndAndModify (string) Required. The collection against which to run the command. query (document) Optional. Species the selection criteria for the modication. The query eld employs the same query selectors (page 763) as used in the db.collection.find() (page 924) method. Although the query may match multiple documents, findAndModify (page 826) will only select one document to modify. sort (document) Optional. Determines which document the operation will modify if the query selects multiple documents. findAndModify (page 826) will modify the rst document in the sort order specied by this argument. remove (boolean) Optional if update eld exists. When true, removes the selected document. The default is false. update (document) Optional if remove eld exists. Performs an update of the selected document. The update eld employs the same update operators (page 784) or field: value specications to modify the selected document. new (boolean) Optional. When true, returns the modied document rather than the original. The findAndModify (page 826) method ignores the new option for remove operations. The default is false. elds (document) Optional. A subset of elds to return. The fields document species an inclusion of a eld with 1, as in the following:
fields: { <field1>: 1, <field2>: 1, ... }
See projection (page 171). upsert (boolean) Optional. Used in conjunction with the update eld. When true, the findAndModify (page 826) command creates a new document if the query returns no documents. The default is false. The findAndModify (page 826) command returns a document, similar to the following:
{ lastErrorObject: { updatedExisting: <boolean>,
826
upserted: <boolean>, n: <num>, connectionId: <num>, err: <string>, ok: <num> } value: <document>, ok: <num> }
The return document contains the following elds: The lastErrorObject eld that returns the details of the command: The updatedExisting eld only appears if the command is either an update or an upsert. The upserted eld only appears if the command is an upsert. The value eld that returns either: the original (i.e. pre-modication) document if new is false, or the modied or inserted document if new: true.
The ok eld that returns the status of the command. Note: If the findAndModify (page 826) nds no matching document, then: for update or remove operations, lastErrorObject does not appear in the return document and the value eld holds a null.
{ "value" : null, "ok" : 1 }
for an upsert operation that performs an insert, when new is false, and includes a sort option, the return document has lastErrorObject, value, and ok elds, but the value eld holds an empty document {}. for an upsert that performs an insert, when new is false without a specied sort the return document has lastErrorObject, value, and ok elds, but the value eld holds a null. Changed in version 2.2: Previously, the command returned an empty document (e.g. {}) in the value eld. See the 2.2 release notes (page 1160) for more information. Consider the following examples: The following command updates an existing document in the people collection where the document matches the query criteria:
db.runCommand( { findAndModify: "people", query: { name: "Tom", state: "active", rating: { $gt: 10 } }, sort: { rating: 1 }, update: { $inc: { score: 1 } } } )
This command performs the following actions: 1.The query nds a document in the people collection where the name eld has the value Tom, the state eld has the value active and the rating eld has a value greater than (page 764) 10.
827
2.The sort orders the results of the query in ascending order. If multiple documents meet the query condition, the command will select for modication the rst document as ordered by this sort. 3.The update increments (page 784) the value of the score eld by 1. 4.The command returns a document with the following elds: The lastErrorObject eld that contains the details of the command, including the eld updatedExisting which is true, and The value eld that contains the original (i.e. pre-modication) document selected for this update:
{ "lastErrorObject" : { "updatedExisting" : true, "n" : 1, "connectionId" : 1, "err" : null, "ok" : 1 }, "value" : { "_id" : ObjectId("50f1d54e9beb36a0f45c6452"), "name" : "Tom", "state" : "active", "rating" : 100, "score" : 5 }, "ok" : 1 }
To return the modied document in the value eld, add the new:true option to the command. If no document match the query condition, the command returns a document that contains null in the value eld:
{ "value" : null, "ok" : 1 }
The mongo (page 1036) shell and many drivers provide a findAndModify() (page 926) helper method. Using the shell helper, this previous operation can take the following form:
db.people.findAndModify( { query: { name: "Tom", state: "active", rating: { $gt: 10 } }, sort: { rating: 1 }, update: { $inc: { score: 1 } } } );
However, the findAndModify() (page 926) shell helper method returns just the unmodied document, or the modied document when new is true.
{ "_id" : ObjectId("50f1d54e9beb36a0f45c6452"), "name" : "Tom", "state" : "active", "rating" : 100, "score" : 5 }
The following findAndModify (page 826) command includes the upsert: a new document if no document matches the query condition:
true option to insert
828
db.runCommand( { findAndModify: "people", query: { name: "Gus", state: "active", rating: 100 }, sort: { rating: 1 }, update: { $inc: { score: 1 } }, upsert: true } )
If the command does not nd a matching document, the command performs an upsert and returns a document with the following elds: The lastErrorObject eld that contains the details of the command, including the eld upserted that contains the ObjectId of the newly inserted document, and The value eld that contains an empty document {} as the original document because the command included the sort option:
{ "lastErrorObject" : { "updatedExisting" : false, "upserted" : ObjectId("50f2329d0092b46dae1dc98e"), "n" : 1, "connectionId" : 1, "err" : null, "ok" : 1 }, "value" : { }, "ok" : 1 }
If the command did not include the sort option, the value eld would contain null:
{ "value" : null, "lastErrorObject" : { "updatedExisting" : false, "n" : 1, "upserted" : ObjectId("5102f7540cb5c8be998c2e99") }, "ok" : 1 }
The following findAndModify (page 826) command includes both upsert: true option and the new:true option to return the newly inserted document in the value eld if a document matching the query is not found:
db.runCommand( { findAndModify: "people", query: { name: "Pascal", state: "active", rating: 25 }, sort: { rating: 1 }, update: { $inc: { score: 1 } }, upsert: true, new: true } )
829
The command returns the newly inserted document in the value eld:
{ "lastErrorObject" : { "updatedExisting" : false, "upserted" : ObjectId("50f47909444c11ac2448a5ce"), "n" : 1, "connectionId" : 1, "err" : null, "ok" : 1 }, "value" : { "_id" : ObjectId("50f47909444c11ac2448a5ce"), "name" : "Pascal", "rating" : 25, "score" : 1, "state" : "active" }, "ok" : 1 }
When the findAndModify (page 826) command includes the upsert: true option and the query eld(s) is not uniquely indexed, the method could insert a document multiple times in certain circumstances. For instance, if multiple clients issue the findAndModify (page 826) command and these commands complete the find phase before any one starts the modify phase, these commands could insert the same document. Consider an example where no document with the name Andy exists and multiple clients issue the following command:
db.runCommand( { findAndModify: "people", query: { name: "Andy" }, sort: { rating: 1 }, update: { $inc: { score: 1 } }, upsert: true } )
If all the commands nish the query phase before any command starts the modify phase, and there is no unique index on the name eld, the commands may all perform an upsert. To prevent this condition, create a unique index (page 334) on the name eld. With the unique index in place, then the multiple findAndModify (page 826) commands would observe one of the following behaviors: Exactly one findAndModify (page 826) would successfully insert a new document. Zero or more findAndModify (page 826) commands would update the newly inserted document. Zero or more findAndModify (page 826) commands would fail when they attempted to insert a duplicate. If the command fails due to a unique index constraint violation, you can retry the command. Absent a delete of the document, the retry should not fail. Warning: When using findAndModify (page 826) in a sharded environment, the query must contain the shard key for all operations against the shard cluster. findAndModify (page 826) operations issued against mongos (page 1032) instances for non-sharded collections function normally.
Note: This command obtains a write lock on the affected database and will block other operations until it has 830 Chapter 63. MongoDB Interface
completed; however, typically the write lock is short lived and equivalent to other similar update() (page 948) operations.
getLastError Denition getLastError The getLastError (page 831) command returns the error status of the last operation on the current connection. By default MongoDB does not provide a response to conrm the success or failure of a write operation, clients typically use getLastError (page 831) in combination with write operations to ensure that the write succeeds. Consider the following prototype form.
{ getLastError: 1 }
The following options are available: Parameters j (boolean) If true, wait for the next journal commit before returning, rather than a full disk ush. If mongod (page 1021) does not have journaling enabled, this option has no effect. If you specify j for a write operation, mongod (page 1021) will wait no more than 1/3 of the current journalCommitInterval (page 1083) before writing data to the journal. w When running with replication, this is the number of servers to replicate to before returning. A w value of 1 indicates the primary only. A w value of 2 includes the primary and at least one secondary, etc. In place of a number, you may also set w to majority to indicate that the command should wait until the latest write propagates to a majority of replica set members. If using w, you should also use wtimeout. Specifying a value for w without also providing a wtimeout may cause getLastError (page 831) to block indenitely. fsync (boolean) If true, wait for mongod (page 1021) to write this data to disk before returning. Defaults to false. In most cases, use the j option to ensure durability and consistency of the data set. wtimeout (integer) Optional. Milliseconds. Specify a value in milliseconds to control how long to wait for write propagation to complete. If replication does not complete in the given timeframe, the getLastError (page 831) command will return with an error status. See also: Write Concern (page 398), Replica Acknowledged (page 399), and db.getLastError() (page 978). Examples Conrm Replication to Two Replica Set Members The following example ensures the operation has replicated to two members (the primary and one other member):
db.runCommand( { getLastError: 1, w: 2 } )
831
Conrm Replication to a Majority of a Replica Set The following example ensures the write operation has replicated to a majority of the congured members of the set.
db.runCommand( { getLastError: 1, w: "majority" } )
Set a Timeout for a getLastError Response Unless you specify a timeout, a getLastError (page 831) command may block forever if MongoDB cannot satisfy the requested write concern. To specify a timeout of 5000 milliseconds, use an invocation that resembles the following:
db.runCommand( { getLastError: 1, w: 2, wtimeout:5000 } )
getPrevError getPrevError The getPrevError (page 832) command returns the errors since the last resetError (page 832) command. See also: db.getPrevError() (page 979) resetError resetError The resetError (page 832) command resets the last error status. See also: db.resetError() (page 984) text Denition text New in version 2.4. The text (page 832) command provides an interface to search text content stored in the text index (page 338). By default, the command limits the matches to the top 100 scoring documents, in descending score order, but you can specify a different limit. The text (page 832) command is case-insensitive. The text (page 832) has the following syntax:
db.collection.runCommand( "text", { search: <string>, filter: <document>, project: <document>, limit: <number>, language: <string> } )
The text (page 832) command has the following parameters: Parameters search (string) A string of terms that MongoDB parses and uses to query the text index. The text (page 832) command returns all documents that contain any of the terms; i.e. it performs a logical OR search. Enclose the string of terms in escaped double quotes to match on the phrase.
832
Additionally, the text (page 832) command treats most punctuation as delimiters, except when a hyphen - is used to negate terms. Prexing a word with a minus sign (-) negates a word: The negated word excludes documents that contain the negated word from the result set. A search string that only contains negated words returns no match. A hyphenated word, such as pre-market, is not a negation. The text command treats the hyphen as a delimiter. lter (document) Optional. A query document (page 190) to further limit the results of the query using another database eld. You can use any valid MongoDB query in the lter document, except if the index includes an ascending or descending index eld as a prex. If the index includes an ascending or descending index eld as a prex, the filter is required and the filter query must be an equality match. project (document) Optional. Allows you to limit the elds returned by the query to only those specied. By default, the _id eld returns as part of the result set unless you explicitly exclude the eld in the project document. limit (number) Optional. Specify the maximum number of documents to include in the response. The text (page 832) sorts the results before applying the limit. The default limit is 100. language (string) Optional. Specify the language that determines for the search the list of stop words and the rules for the stemmer and tokenizer. If not specied, the search uses the default language of the index. See Text Search Languages (page 836) for the supported languages. Specify the language in lowercase. Returns A document that contains a eld results that contains an array of the highest scoring documents, in descending order by score. See Output (page 836) for details. The returned document must t within the BSON Document Size (page 1105). Otherwise, the command will return as many results as not to exceed the BSON Document Size (page 1105). Use the limit and the project parameters to limit the size of the result set. Note: If the search string includes phrases, the search performs an AND with any other terms in the search string; e.g. search for "\"twinkle twinkle\" little star" searches for "twinkle twinkle" and ("little" or "star"). text (page 832) adds all negations to the query with the logical AND operator. The text (page 832) command ignores stop words for the search language, such as the and and in English. The text (page 832) command matches on the complete stemmed word. So if a document eld contains the word blueberry, a search on the term blue will not match. However, blueberry or blueberries will match. For the following examples, assume a collection articles has a text index on the eld subject:
db.articles.ensureIndex( { subject: "text" } )
833
Example Search for a Single Word

db.articles.runCommand( "text", { search: "coffee" } )
This query returns documents that contain the word coffee, case-insensitive, in the indexed subject eld. Example Search for Multiple Words The following command searches for bake or coffee or cake:
db.articles.runCommand( "text", { search: "bake coffee cake" } )
This query returns documents that contain either bake or coffee or cake in the indexed subject eld. Example Search for a Phrase
db.articles.runCommand( "text", { search: "\"bake coffee cake\"" } )
This query returns documents that contain the phrase bake coffee cake. Example Exclude a Term from the Result Set Use the hyphen (-) as a prex to exclude documents that contain a term. Search for documents that contain the words bake or coffee but do not contain cake:
db.articles.runCommand( "text", { search: "bake coffee -cake" } )
Example Search with Additional Query Conditions Use the filter option to include additional query conditions. Search for a single word coffee with an additional lter on the about eld, but limit the results to 2 documents with the highest score and return only the subject eld in the matching documents:
db.articles.runCommand( "text", { search: "coffee", filter: { about: /desserts/ }, limit: 2, project: { subject: 1, _id: 0 } } )
The filter query document (page 190) may use any of the available query operators (page 763). Because the _id eld is implicitly included, in order to return only the subject eld, you must explicitly exclude (0) the _id eld. Within the project document, you cannot mix inclusions (i.e. <fieldA>: 1) and exclusions (i.e. <fieldB>: 0), except for the _id eld.
834
Example Search a Different Language Use the language option to specify Spanish as the language that determines the list of stop words and the rules for the stemmer and tokenizer:
db.articles.runCommand( "text", { search: "leche", language: "spanish" } )
See Text Search Languages (page 836) for the supported languages. Important: Specify the language in lowercase.
Example The following is an example document returned by the text (page 832) command: Warning: The complete results of the text (page 832) command must t within the BSON Document Size (page 1105). Otherwise, the command will limit the results to t within the BSON Document Size (page 1105). Use the limit and the project parameters with the text (page 832) command to limit the size of the result set.
{ "queryDebugString" : "tomorrow||||||", "language" : "english", "results" : [ { "score" : 1.3125, "obj": { "_id" : ObjectId("50ecef5f8abea0fda30ceab3"), "quote" : "tomorrow, and tomorrow, and tomorrow, creeps in this petty pace", "related_quotes" : [ "is this a dagger which I see before me", "the handle toward my hand?" ], "src" : { "title" : "Macbeth", "from" : "Act V, Scene V" }, "speaker" : "macbeth" } } ], "stats" : { "nscanned" : 1, "nscannedObjects" : 0, "n" : 1, "nfound" : 1, "timeMicros" : 163 }, "ok" : 1 }
835
Output text.queryDebugString For internal use only. text.language The language (page 836) eld returns the language used for the text search. This language determines the list of stop words and the rules for the stemmer and tokenizer. text.results The results (page 836) eld returns an array of result documents that contain the information on the matching documents. The result documents are ordered by the score (page 836). Each result document contains: text.results.obj The obj (page 836) eld returns the actual document from the collection that contained the stemmed term or terms. text.results.score The score (page 836) eld for the document that contained the stemmed term or terms. The score (page 836) eld signies how well the document matched the stemmed term or terms. See Control Results of Text Search with Weights (page 375) for how you can adjust the scores for the matching words. text.stats The stats (page 836) eld returns a document that contains the query execution statistics. The stats (page 836) eld contains: text.stats.nscanned The nscanned (page 836) eld returns the total number of index entries scanned. text.stats.nscannedObjects The nscannedObjects (page 836) eld returns the total number of documents scanned. text.stats.n The n (page 836) eld returns the number of elements in the results (page 836) array. This number may be less than the total number of matching documents, i.e. nfound (page 836), if the full result exceeds the BSON Document Size (page 1105). text.stats.nfound The nfound (page 836) eld returns the total number number of documents that match. This number may be greater than the size of the results (page 836) array, i.e. n (page 836), if the result set exceeds the BSON Document Size (page 1105). text.stats.timeMicros The timeMicros (page 836) eld returns the time in microseconds for the search. text.ok The ok (page 836) returns the status of the text (page 832) command. Text Search Languages guages: danish dutch english finnish french german hungarian 836 Chapter 63. MongoDB Interface The text index (page 338) and the text (page 832) command support the following lan-
italian norwegian portuguese romanian russian spanish swedish turkish Note: If you specify a language value of "none", then the text search has no list of stop words, and the text search does not stem or tokenize the search terms.
63.2.2 Database Operations

Replication Commands
Replication Commands
Name applyOps (page 837) getoptime (page 838) isMaster (page 838) replSetFreeze (page 840) replSetGetStatus (page 840) replSetInitiate (page 842) replSetMaintenance (page 843) replSetReconfig (page 844) replSetStepDown (page 844) replSetSyncFrom (page 845) resync (page 846)
Description Internal command that applies applies oplog entries to the current data set. Internal command to support replication, returns the optime. Displays information about this members role in the replica set, including whether it is the master. Prevents the current member from seeking election as primary for a period of time. Returns a document that reports on the status of the replica set. Initializes a new replica set. Enables or disables a maintenance mode, which puts a secondary node in a RECOVERING state. Applies a new conguration to an existing replica set. Forces the current primary to step down and become a secondary, forcing an election. Explicitly override the default logic for selecting a member to replicate from. Forces a mongod (page 1021) to re-synchronize from the master. For master-slave replication only.
applyOps applyOps Parameters operations (array) an array of operations to perform. preCondition (array) Optional. Denes one or more conditions that the destination must meet applying the entries from the <operations> array. Use ns to specify a namespace, 63.2. Database Commands 837
q to specify a query and res to specify the result that the query should match. You may specify zero, one, or many preCondition documents. applyOps (page 837) provides a way to apply entries from an oplog created by replica set members and master instances in a master/slave deployment. applyOps (page 837) is primarily an internal command to support sharding functionality, and has the following prototype form:
db.runCommand( { applyOps: [ <operations> ], preCondition: [ { ns: <namespace>, q: <query>, res:
applyOps (page 837) applies oplog entries from the <operations> array, to the mongod (page 1021) instance. The preCondition array provides the ability to specify conditions that must be true in order to apply the oplog entry. You can specify as many preCondition sets as needed. If you specify the ns option, applyOps (page 837) will only apply oplog entries for the collection described by that namespace. You may also specify a query in the q eld with a corresponding expected result in the res eld that must match in order to apply the oplog entry. Warning: This command obtains a global write lock and will block other operations until it has completed.
getoptime getoptime getoptime (page 838) is an internal command. isMaster Denition isMaster isMaster (page 838) returns a document that describes the role of the mongod (page 1021) instance. If the instance is a member of a replica set, then isMaster (page 838) returns a subset of the replica set conguration and status including whether or not the instance is the primary of the replica set. When sent to a mongod (page 1021) instance that is not a member of a replica set, isMaster (page 838) returns a subset of this information. MongoDB drivers and clients use isMaster (page 838) to determine the state of the replica set members and to discover additional members of a replica set. The db.isMaster() (page 981) method in the mongo (page 1036) shell provides a wrapper around isMaster (page 838). The command takes the following form:
{ isMaster: 1 }
See also: db.isMaster() (page 981) Output
838
All Instances The following isMaster (page 838) elds are common across all roles: isMaster.ismaster A boolean value that reports when this node is writable. If true, then this instance is a primary in a replica set, or a master in a master-slave conguration, or a mongos (page 1032) instance, or a standalone mongod (page 1021). This eld will be false if the instance is a secondary member of a replica set or if the member is an arbiter of a replica set. isMaster.maxBsonObjectSize The maximum permitted size of a BSON object in bytes for this mongod (page 1021) process. If not provided, clients should assume a max size of 4 * 1024 * 1024. isMaster.maxMessageSizeBytes New in version 2.4. The maximum permitted size of a BSON wire protocol message. The default value is 48000000 bytes. isMaster.localTime New in version 2.2. Returns the local server time in UTC. This value is an ISO date. Sharded Instances document: mongos (page 1032) instances add the following eld to the isMaster (page 838) response
isMaster.msg Contains the value isdbgrid when isMaster (page 838) returns from a mongos (page 1032) instance. Replica Sets isMaster (page 838) contains these elds when returned by a member of a replica set: isMaster.setName The name of the current :replica set. isMaster.secondary A boolean value that, when true, indicates if the mongod (page 1021) is a secondary member of a replica set. isMaster.hosts An array of strings in the format of "[hostname]:[port]" that lists all members of the replica set that are neither hidden, passive, nor arbiters. Drivers use this array and the isMaster.passives (page 839) to determine which members to read from. isMaster.passives An array of strings in the format of "[hostname]:[port]" listing all members of the replica set which have a priority (page 390) of 0. This eld only appears if there is at least one member with a priority (page 390) of 0. Drivers use this array and the isMaster.hosts (page 839) to determine which members to read from. isMaster.arbiters An array of strings in the format of "[hostname]:[port]" listing all members of the replica set that are arbiters. This eld only appears if there is at least one arbiter in the replica set. isMaster.primary A string in the format of "[hostname]:[port]" listing the current primary member of the replica set.
839
isMaster.arbiterOnly A boolean value that , when true, indicates that the current instance is an arbiter. The arbiterOnly (page 839) eld is only present, if the instance is an arbiter. isMaster.passive A boolean value that, when true, indicates that the current instance is hidden. The passive (page 840) eld is only present for hidden members. isMaster.hidden A boolean value that, when true, indicates that the current instance is hidden. The hidden (page 840) eld is only present for hidden members. isMaster.tags A document that lists any tags assigned to this member. This eld is only present if there are tags assigned to the member. See Congure Replica Set Tag Sets (page 455) for more information. isMaster.me The [hostname]:[port] of the member that returned isMaster (page 838). replSetFreeze replSetFreeze The replSetFreeze (page 840) command prevents a replica set member from seeking election for the specied number of seconds. Use this command in conjunction with the replSetStepDown (page 844) command to make a different node in the replica set a primary. The replSetFreeze (page 840) command uses the following syntax:
{ replSetFreeze: <seconds> }
If you want to unfreeze a replica set member before the specied number of seconds has elapsed, you can issue the command with a seconds value of 0:
{ replSetFreeze: 0 }
Restarting the mongod (page 1021) process also unfreezes a replica set member. replSetFreeze (page 840) is an administrative command, and you must issue it against the admin database. replSetGetStatus Denition replSetGetStatus The replSetGetStatus command returns the status of the replica set from the point of view of the current server. You must run the command against the admin database. The command has the following prototype format:
{ replSetGetStatus: 1 }
The value specied does not affect the output of the command. Data provided by this command derives from data included in heartbeats sent to the current instance by other members of the replica set. Because of the frequency of heartbeats, these data can be several seconds out of date. You can also access this functionality through the rs.status() (page 990) helper in the mongo (page 1036) shell. The mongod (page 1021) must have replication enabled and be a member of a replica set for the for replSetGetStatus (page 840) to return successfully.
840
Output replSetGetStatus.set The set value is the name of the replica set, congured in the replSet (page 1088) setting. This is the same value as _id (page 463) in rs.conf() (page 988). replSetGetStatus.date The value of the date eld is an ISODate of the current time, according to the current server. Compare this to the value of the lastHeartbeat (page 842) to nd the operational lag between the current host and the other hosts in the set. replSetGetStatus.myState The value of myState (page 841) reects state of the current replica set member. An integer between 0 and 10 represents the state of the member. These integers map to states, as described in the following table: Number 0 1 2 3 4 5 6 7 8 9 10 Name STARTUP PRIMARY SECONDARY RECOVERING FATAL STARTUP2 UNKNOWN ARBITER DOWN ROLLBACK SHUNNED State Start up, phase 1 (parsing conguration.) Primary. Secondary. Member is recovering (initial sync, post-rollback, stale members.) Member has encountered an unrecoverable error. Start up, phase 2 (forking threads.) Unknown (the set has never connected to the member.) Member is an arbiter. Member is not accessible to the set. Member is rolling back data. See rollback. Member has been removed from replica set.
replSetGetStatus.members The members eld holds an array that contains a document for every member in the replica set. replSetGetStatus.members.name The name eld holds the name of the server. replSetGetStatus.members.self The self eld is only included in the document for the current mongod instance in the members array. Its value is true. replSetGetStatus.members.errmsg This eld contains the most recent error or status message received from the member. This eld may be empty (e.g. "") in some cases. replSetGetStatus.members.health The health value is only present for the other members of the replica set (i.e. not the member that returns rs.status (page 990).) This eld conveys if the member is up (i.e. 1) or down (i.e. 0.) replSetGetStatus.members.state The value of the state (page 841) reects state of this replica set member. An integer between 0 and 10 represents the state of the member. These integers map to states, as described in the following table:
841
Number 0 1 2 3 4 5 6 7 8 9 10
Name STARTUP PRIMARY SECONDARY RECOVERING FATAL STARTUP2 UNKNOWN ARBITER DOWN ROLLBACK SHUNNED
State Start up, phase 1 (parsing conguration.) Primary. Secondary. Member is recovering (initial sync, post-rollback, stale members.) Member has encountered an unrecoverable error. Start up, phase 2 (forking threads.) Unknown (the set has never connected to the member.) Member is an arbiter. Member is not accessible to the set. Member is rolling back data. See rollback. Member has been removed from replica set.
replSetGetStatus.members.stateStr A string that describes state (page 841). replSetGetStatus.members.uptime The uptime (page 842) eld holds a value that reects the number of seconds that this member has been online. This value does not appear for the member that returns the rs.status() (page 990) data. replSetGetStatus.members.optime A document that contains information regarding the last operation from the operation log that this member has applied. replSetGetStatus.members.optime.t A 32-bit timestamp of the last operation applied to this member of the replica set from the oplog. replSetGetStatus.members.optime.i An incremented eld, which reects the number of operations in since the last time stamp. This value only increases if there is more than one operation per second. replSetGetStatus.members.optimeDate An ISODate formatted date string that reects the last entry from the oplog that this member applied. If this differs signicantly from lastHeartbeat (page 842) this member is either experiencing replication lag or there have not been any new operations since the last update. Compare members.optimeDate between all of the members of the set. replSetGetStatus.members.lastHeartbeat The lastHeartbeat value provides an ISODate formatted date of the last heartbeat received from this member. Compare this value to the value of the date (page 841) eld to track latency between these members. This value does not appear for the member that returns the rs.status() (page 990) data. replSetGetStatus.members.pingMS The pingMS represents the number of milliseconds (ms) that a round-trip packet takes to travel between the remote member and the local instance. This value does not appear for the member that returns the rs.status() (page 990) data. replSetGetStatus.syncingTo The syncingTo eld is only present on the output of rs.status() (page 990) on secondary and recovering members, and holds the hostname of the member from which this instance is syncing. replSetInitiate replSetInitiate The replSetInitiate (page 842) command initializes a new replica set. Use the following syntax:
842
{ replSetInitiate : <config_document> }
The <config_document> is a document that species the replica sets conguration. For instance, heres a cong document for creating a simple 3-member replica set:
A typical way of running this command is to assign the cong document to a variable and then to pass the document to the rs.initiate() (page 989) helper:
config = { _id : "my_replica_set", members : [ {_id : 0, host : "rs1.example.net:27017"}, {_id : 1, host : "rs2.example.net:27017"}, {_id : 2, host : "rs3.example.net", arbiterOnly: true}, ] } rs.initiate(config)
Notice that omitting the port cause the host to use the default port of 27017. Notice also that you can specify other options in the cong documents such as the arbiterOnly setting in this example. See also: Replica Set Conguration (page 463), Replica Set Administration (page 419), and Replica Set Reconguration (page 467). replSetMaintenance replSetMaintenance The replSetMaintenance (page 843) admin command enables or disables the maintenance mode for a secondary member of a replica set. The command has the following prototype form:
{ replSetMaintenance: <boolean> }
Consider the following behavior when running the replSetMaintenance (page 843) command: You cannot run the command on the Primary. You must run the command against the admin database. When enabled replSetMaintenance: secondary is RECOVERING: 1, the member enters the RECOVERING state. While the
The member is not accessible for read operations. The member continues to sync its oplog from the Primary.
843
replSetRecong replSetReconfig The replSetReconfig (page 844) command modies the conguration of an existing replica set. You can use this command to add and remove members, and to alter the options set on existing members. Use the following syntax:
{ replSetReconfig: <new_config_document>, force: false }
You may also run the command using the shells rs.reconfig() (page 989) method. Be aware of the following replSetReconfig (page 844) behaviors: You must issue this command against the admin database of the current primary member of the replica set. You can optionally force the replica set to accept the new conguration by specifying force: true. Use this option if the current member is not primary or if a majority of the members of the set are not accessible. Warning: Forcing the replSetReconfig (page 844) command can lead to a rollback situation. Use with caution. Use the force option to restore a replica set to new servers with different hostnames. This works even if the set members already have a copy of the data. A majority of the sets members must be operational for the changes to propagate properly. This command can cause downtime as the set renegotiates primary-status. Typically this is 10-20 seconds, but could be as long as a minute or more. Therefore, you should attempt to recongure only during scheduled maintenance periods. In some cases, replSetReconfig (page 844) forces the current primary to step down, initiating an election for primary among the members of the replica set. When this happens, the set will drop all current connections. Note: replSetReconfig (page 844) obtains a special mutually exclusive lock to prevent more than one replSetReconfig (page 844) operation from occurring at the same time.
replSetStepDown replSetStepDown Options force (boolean) Forces the primary to step down even if there arent any secondary members within 10 seconds of the primarys latest optime. This option is not available in versions of mongod (page 1021) before 2.0. The replSetStepDown (page 844) command forces the primary of the replica set to relinquish its status as primary. This initiates an election for primary (page 409). You may specify a number of seconds for the node to avoid election to primary:
{ replSetStepDown: <seconds> }
If you do not specify a value for <seconds>, replSetStepDown (page 844) will attempt to avoid reelection to primary for 60 seconds. Warning: This will force all clients currently connected to the database to disconnect. This help to ensure that clients maintain an accurate view of the replica set.
844
New in version 2.0: If there is no secondary, within 10 seconds of the primary, replSetStepDown (page 844) will not succeed to prevent long running elections. replSetSyncFrom replSetSyncFrom New in version 2.2. Options host Species the name and port number of the replica set member that this member replicates from. Use the [hostname]:[port] form. replSetSyncFrom (page 845) allows you to explicitly congure which host the current mongod (page 1021) will poll oplog entries from. This operation may be useful for testing different patterns and in situations where a set member is not replicating from the host you want. The member to replicate from must be a valid source for data in the set. A member cannot replicate from: itself. an arbiter, because arbiters do not hold data. a member that does not build indexes. an unreachable member. a mongod (page 1021) instance that is not a member of the same replica set. If you attempt to replicate from a member that is more than 10 seconds behind the current member, mongod (page 1021) will return and log a warning, but it still will replicate from the member that is behind. If you run rs.syncFrom() (page 991) during initial sync, MongoDB produces no error messages, but the sync target will not change until after the initial sync operation. The command has the following prototype form:
{ replSetSyncFrom: "[hostname]:[port]" }
To run the command in the mongo (page 1036) shell, use the following invocation:
db.adminCommand( { replSetSyncFrom: "[hostname]:[port]" } )
You may also use the rs.syncFrom() (page 991) helper in the mongo (page 1036) shell, in an operation with the following form:
rs.syncFrom("[hostname]:[port]")
Note: replSetSyncFrom (page 845) and rs.syncFrom() (page 991) provide a temporary override of default behavior. If: the mongod (page 1021) instance restarts, the connection to the sync target closes, or Changed in version 2.4: The sync target falls more than 30 seconds behind another member of the replica set; then, the mongod (page 1021) instance will revert to the default sync logic and target.
845
resync resync The resync (page 846) command forces an out-of-date slave mongod (page 1021) instance to re-synchronize itself. Note that this command is relevant to master-slave replication only. It does not apply to replica sets. Warning: This command obtains a global write lock and will block other operations until it has completed. See also: Replica Set Fundamental Concepts (page 387) for more information regarding replication. Sharding Commands
Sharding Commands
Name addShard (page 846) checkShardingIndex (page 848) enableSharding (page 848) flushRouterConfig (page 848) getShardMap (page 848) getShardVersion (page 848) isdbgrid (page 848) listShards (page 849) medianKey (page 849) moveChunk (page 849) movePrimary (page 850) removeShard (page 850) setShardVersion (page 851) shardCollection (page 851) shardingState (page 852) splitChunk (page 852) splitVector (page 853) split (page 853) unsetSharding (page 855) addShard addShard
Description Adds a shard to a sharded cluster. Internal command that validates index on shard key. Enables sharding on a specic database. Forces an update to the cluster metadata cached by a mongos (page 1032). Internal command that reports on the state of a sharded cluster. Internal command that returns the cong server version. Veries that a process is a mongos (page 1032). Returns a list of congured shards. Deprecated internal command. See splitVector (page 853). Internal command that migrates chunks between shards. Reassigns the primary shard when removing a shard from a sharded cluster. Starts the process of removing a shard from a sharded cluster. Internal command to sets the cong server version. Enables the sharding functionality for a collection, allowing the collection to be sharded. Reports whether the mongod (page 1021) is a member of a sharded cluster. Internal command to split chunk. Instead use the methods sh.splitFind() (page 1001) and sh.splitAt() (page 1001). Internal command that determines split points. Creates a new chunk. Internal command that affects connections between instances in a MongoDB deployment.
846
Parameters hostname (string) a hostname or replica-set/hostname string. name (string) Optional. Unless specied, a name will be automatically provided to uniquely identify the shard. maxSize (integer) Optional, megabytes. Limits the maximum size of a shard. If maxSize is 0 then MongoDB will not limit the size of the shard. Use the addShard (page 846) command to add a database instance or replica set to a sharded cluster. You must run this command when connected a mongos (page 1032) instance. The command takes the following form:
{ addShard: "<hostname><:port>" }
Example
db.runCommand({addShard: "mongodb0.example.net:27027"})
Replace <hostname><:port> with the hostname and port of the database instance you want to add as a shard. Warning: Do not use localhost for the hostname unless your conguration server is also running on localhost. The optimal conguration is to deploy shards across replica sets. To add a shard on a replica set you must specify the name of the replica set and the hostname of at least one member of the replica set. You must specify at least one member of the set, but can specify all members in the set or another subset if desired. addShard (page 846) takes the following form:
{ addShard: "replica-set/hostname:port" }
Example
db.runCommand( { addShard: "repl0/mongodb3.example.net:27327"} )
If you specify additional hostnames, all must be members of the same replica set. Send this command to only one mongos (page 1032) instance, it will store shard conguration information in the cong database. Note: Specify a maxSize when you have machines with different disk capacities, or if you want to limit the amount of data on some shards. The maxSize constraint prevents the balancer from migrating chunks to the shard when the value of mem.mapped (page 894) exceeds the value of maxSize. See also: sh.addShard() (page 995) Sharded Cluster Administration (page 503) Add Shards to a Cluster (page 510) Remove Shards from an Existing Sharded Cluster (page 530)
847
checkShardingIndex checkShardingIndex checkShardingIndex (page 848) is an internal command that supports the sharding functionality. enableSharding enableSharding The enableSharding (page 848) command enables sharding on a per-database level. Use the following command form:
{ enableSharding: "<database name>" }
Once youve enabled sharding in a database, you can use the shardCollection (page 851) command to begin the process of distributing data among the shards. ushRouterCong flushRouterConfig flushRouterConfig (page 848) clears the current cluster information cached by a mongos (page 1032) instance and reloads all sharded cluster metadata from the cong database. This forces an update when the conguration database holds data that is newer than the data cached in the mongos (page 1032) process. Warning: Do not modify the cong data, except as explicitly documented. A cong database cannot typically tolerate manual manipulation. flushRouterConfig (page 848) is an administrative command that is only available for mongos (page 1032) instances. New in version 1.8.2. getShardMap getShardMap getShardMap (page 848) is an internal command that supports the sharding functionality. getShardVersion getShardVersion getShardVersion (page 848) is an internal command that supports sharding functionality. isdbgrid isdbgrid This command veries that a process is a mongos (page 1032). If you issue the isdbgrid (page 848) command when connected to a mongos (page 1032), the response document includes the isdbgrid eld set to 1. The returned document is similar to the following:
{ "isdbgrid" : 1, "hostname" : "app.example.net", "ok" : 1 }
If you issue the isdbgrid (page 848) command when connected to a mongod (page 1021), MongoDB returns an error document. The isdbgrid (page 848) command is not available to mongod (page 1021). The error document, however, also includes a line that reads "isdbgrid" : 1, just as in the document returned for a mongos (page 1032). The error document is similar to the following:
848
{ "errmsg" : "no such cmd: isdbgrid", "bad cmd" : { "isdbgrid" : 1 }, "ok" : 0 }
You can instead use the isMaster (page 838) command to determine connection to a mongos (page 1032). When connected to a mongos (page 1032), the isMaster (page 838) command returns a document that contains the string isdbgrid in the msg eld. listShards listShards Use the listShards (page 849) command to return a list of congured shards. The command takes the following form:
{ listShards: 1 }
medianKey medianKey medianKey (page 849) is an internal command. moveChunk moveChunk moveChunk (page 849) is an internal administrative command that moves chunks between shards. You must issue the moveChunk (page 849) command via a mongos (page 1032) instance while using the admin database in the following form:
db.runCommand( { moveChunk : <namespace> , find : <query> , to : <destination>, <options> } )
Parameters moveChunk (string) The name of the collection where the chunk exists. Specify the collections full namespace, including the database name. nd (document) A document that species an equality match on the shard key that will move the chunk that contains the specied shard-key value. You may specify either bounds or find but not both. bounds (array) Specify the bounds of a specic chunk to move. The array must consist of two documents specifying the lower and upper shard key values of a chunk to move. You may specify either bounds or find but not both. Use bounds to move chunks in collections partitioned using a hashed shard key. to (string) The name of the destination shard for the chunk. _secondaryThrottle (boolean) Optional; true by default. When true, the balancer waits for replication to secondaries while copying and deleting data during migrations. For details, see Require Replication before Chunk Migration (Secondary Throttle) (page 527).
849
The chunk migration (page 499) section describes how chunks move between shards on MongoDB. moveChunk (page 849) returns the following message if another metadata operation is in progress chunks (page 547) collection:
errmsg: "The collections metadata lock is already taken."
If another process, such as a balancer process, changes meta data while moveChunk (page 849) is running, you may see this error. You may retry the moveChunk (page 849) operation without side effects. Note: Only use the moveChunk (page 849) in special circumstances such as preparing your sharded cluster for an initial ingestion of data, or a large bulk import operation. In most cases allow the balancer to create and balance chunks in sharded clusters. See Create Chunks (Pre-Splitting) (page 522) for more information. See also: split (page 853), sh.moveChunk() (page 999), sh.splitAt() (page 1001), and sh.splitFind() (page 1001). movePrimary movePrimary In a sharded cluster, this command reassigns the databases primary shard, which holds all un-sharded collections in the database. movePrimary (page 850) is an administrative command that is only available for mongos (page 1032) instances. Only use movePrimary (page 850) when removing a shard from a sharded cluster. Important: Only use movePrimary (page 850) when: the database does not contain any collections with data, or you have drained all sharded collections using the removeShard (page 850) command. See Remove Shards from an Existing Sharded Cluster (page 530) for a complete procedure. movePrimary (page 850) changes the primary shard for this database in the cluster metadata, and migrates all un-sharded collections to the specied shard. Use the command with the following form:
{ movePrimary : "test", to : "shard0001" }
When the command returns, the databases primary location will shift to the designated shard. To fully decommission a shard, use the removeShard (page 850) command. removeShard removeShard Starts the process of removing a shard from a cluster. This is a multi-stage process. Begin by issuing the following command:
{ removeShard : "[shardName]" }
The balancer will then migrate chunks from the shard specied by [shardName]. This process happens slowly to avoid placing undue load on the overall cluster. The command returns immediately, with the following message:
{ msg : "draining started successfully" , state: "started" , shard: "shardName" , ok : 1 }
If you run the command again, youll see the following progress output:
850
{ msg: "draining ongoing" , state: "ongoing" , remaining: { chunks: 23 , dbs: 1 }, ok: 1 }
The remaining document species how many chunks and databases remain on the shard. db.printShardingStatus() (page 983) to list the databases that you must move from the shard.
Use
Each database in a sharded cluster has a primary shard. If the shard you want to remove is also the primary of one of the clusters databases, then you must manually move the database to a new shard. This can be only after the shard is empty. See the movePrimary (page 850) command for details. After removing all chunks and databases from the shard, you may issue the command again, to return:
{ msg: "remove shard completed successfully", stage: "completed", host: "shardName", ok : 1 }
setShardVersion setShardVersion setShardVersion (page 851) is an internal command that supports sharding functionality. shardCollection shardCollection The shardCollection (page 851) command marks a collection for sharding and will allow data to begin distributing among shards. You must run enableSharding (page 848) on a database before running the shardCollection (page 851) command.
{ shardCollection: "<database>.<collection>", key: <shardkey> }
This enables sharding for the collection specied by <collection> in the database named <database>, using the key <shardkey> to distribute documents among the shard. <shardkey> is a document and takes the same form as an index specication document (page 192). Parameters shardCollection (string) Specify the namespace of a collection to shard in the form <database>.<collection>. key (document) Specify the index specication to use as the shard key. The index must exist prior to the shardCollection (page 851) command unless the collection is empty. If the collection is empty, then MongoDB will create the index prior to sharding the collection. New in version 2.4: The key may be in the form { field : use the specied eld as a hashed shard key (page 496) . "hashed" } which will
unique (boolean) When true, the unique option ensures that the underlying index enforces a unique constraint. Hashed shard keys do not support unique constraints. numInitialChunks (integer) New in version 2.4. Specify the number of chunks to create upon sharding the collection. The collection will then be pre-split and balanced across the specied number of chunks. You can specify no more than 8192 chunks using numInitialChunks. Choosing the right shard key to effectively distribute load among your shards requires some planning. Also review Shard Keys (page 485) regarding choosing a shard key. Warning: MongoDB provides no method to deactivate sharding for a collection after calling shardCollection (page 851). Additionally, after shardCollection (page 851), you cannot change shard keys or modify the value of any eld used in your shard key index.
851
See also: Sharding (page 483), Sharded Cluster Overview (page 485), and Deploy a Sharded Cluster (page 503). shardingState shardingState shardingState (page 852) is an admin command that reports if mongod (page 1021) is a member of a sharded cluster. shardingState (page 852) has the following prototype form:
{ shardingState: 1 }
For shardingState (page 852) to detect that a mongod (page 1021) is a member of a sharded cluster, the mongod (page 1021) must satisfy the following conditions: 1.the mongod (page 1021) is a primary member of a replica set, and 2.the mongod (page 1021) instance is a member of a sharded cluster. If shardingState (page 852) detects that a mongod (page 1021) is a member of a sharded cluster, shardingState (page 852) returns a document that resembles the following prototype:
{ "enabled" : true, "configServer" : "<configdb-string>", "shardName" : "<string>", "shardHost" : "string:", "versions" : { "<database>.<collection>" : Timestamp(<...>), "<database>.<collection>" : Timestamp(<...>) }, "ok" : 1 }
Otherwise, shardingState (page 852) will return the following document:

{ "note" : "from execCommand", "ok" : 0, "errmsg" : "not master" }
The response from shardingState (page 852) when used with a cong server is:
{ "enabled": false, "ok": 1 }
Note: mongos (page 1032) instances do not provide the shardingState (page 852). Warning: This command obtains a write lock on the affected database and will block other operations until it has completed; however, the operation is typically short lived.
splitChunk splitChunk splitChunk (page 852) is an internal administrative command. Use the sh.splitFind() (page 1001) and sh.splitAt() (page 1001) functions in the mongo (page 1036) shell to split chunks.
852
Warning: Be careful when splitting data in a sharded collection to create new chunks. When you shard a collection that has existing data, MongoDB automatically creates chunks to evenly distribute the collection. To split data effectively in a sharded cluster you must consider the number of documents in a chunk and the average document size to create a uniform chunk size. When chunks have irregular sizes, shards may have an equal number of chunks but have very different data sizes. Avoid creating splits that lead to a collection with differently sized chunks. Parameters ns (string) A string with the complete namespace of the chunk to split. keyPattern (document) Species the shard key. min (document) Species the lower bound of the shard key for the chunk to split. max (document) Species the upper bound of the shard key for the chunk to split. from (string) Species the shard that owns the chunk to split. splitKeys (document) Species the split point for the chunk. shardId (document) Species the shard. See also: moveChunk (page 849), sh.moveChunk() (page 999), sh.splitFind() (page 1001). sh.splitAt() (page 1001), and
splitVector splitVector Is an internal command that supports meta-data operations in sharded clusters. split split The split (page 853) command creates new chunks in a sharded environment. While splitting is typically managed automatically by the mongos (page 1032) instances, this command makes it possible for administrators to manually create splits. In most clusters, MongoDB will manage all chunk creation and distribution operations without manual intervention. You must issue the split (page 853) command while connected to the admin database of a mongos (page 1032) instance.
db.runCommand( { split: <database>.<collection>, find: <document> } )
Or:
db.runCommand( { split: <database>.<collection>, middle: <document> } )
Or:
db.runCommand( { split: <database>.<collection>, bounds: [ lower, upper ] } )
Parameters split (string) The name of the collection where the chunk exists. Specify the collections full namespace, including the database name.
853
nd (document) A document that species an equality match on the shard key that will select the chunk that contains the specied document. You must specify only one of: find, bounds, or middle. bounds (array) New in version 2.4. Specify the bounds of a specic chunk to split in half. The array must consist of two documents specifying the lower and upper shard key values of a chunk to split. The values must match the minimum and maximum values of an existing chunk. You must specify only one of: find, bounds, or middle. Use bounds to move chunks in collections partitioned using a hashed shard key. middle (document) Specify a document to use as the split point to create two chunks. split (page 853) requires one of the following options: find, bounds, or middle. If you need to create a split for a collection that uses a hashed shard key use the bounds parameter with split (page 853), not middle. Warning: Be careful when splitting data in a sharded collection to create new chunks. When you shard a collection that has existing data, MongoDB automatically creates chunks to evenly distribute the collection. To split data effectively in a sharded cluster you must consider the number of documents in a chunk and the average document size to create a uniform chunk size. When chunks have irregular sizes, shards may have an equal number of chunks but have very different data sizes. Avoid creating splits that lead to a collection with differently sized chunks.
Example
db.runCommand( { split : "test.people" , find : { _id : 99 } } )
This command inserts a new split in the collection named people in the test database. This will split the chunk that contains the document that matches the query { _id : 99 } in half. If the document specied by the query does not (yet) exist, the split (page 853) will divide the chunk where that document would exist. The split divides the chunk in half, and does not split the chunk using the identied document as the middle. Example To dene an arbitrary split point, use the following form:
db.runCommand( { split : "test.people" , middle : { _id : 99 } } )
This form is typically used when pre-splitting data in a collection. Example To split a specic chunk using the minimum and maximum values of the hashed shard key of that chunk use the following:
db.runCommand( { split: "test.people" , bounds : [ NumberLong("-5838464104018346494"), NumberLong("-5557153028469814163")] } )
The chunk migration (page 499) section describes how chunks move between shards on MongoDB. If another process, such as a balancer process, changes meta data while split (page 853) is running, you may see this error. You may retry the split (page 853) operation without side effects.
854
errmsg: "The collections metadata lock is already taken."
split (page 853) is an administrative command that is only available for mongos (page 1032) instances. If another process, such as a balancer process, changes meta data while split (page 853) is running, you may see this error. You may retry the split (page 853) operation without side effects. See also: moveChunk (page 849), sh.moveChunk() (page 999), sh.splitFind() (page 1001). sh.splitAt() (page 1001), and
unsetSharding unsetSharding unsetSharding (page 855) is an internal command that supports sharding functionality. See also: Sharding (page 483) for more information about MongoDBs sharding functionality. Instance Administration Commands
Administration Commands
Name clean (page 856) cloneCollectionAsCapped (page 856) cloneCollection (page 856) clone (page 857) closeAllDatabases (page 857) collMod (page 857) compact (page 859) connPoolSync (page 861) convertToCapped (page 861) copydb (page 862) create (page 863) dropDatabase (page 863) dropIndexes (page 864) drop (page 864) filemd5 (page 864) fsync (page 865) getParameter (page 866) logRotate (page 866) reIndex (page 867) renameCollection (page 867) repairDatabase (page 868) setParameter (page 869) shutdown (page 869) touch (page 870)
Description Internal namespace administration command. Copies a non-capped collection as a new capped collection. Copies a collection from a remote host to the current host. Copies a database from a remote host to the current host. Internal command that invalidates all cursors and closes open database les. Add ags to collection to modify the behavior of MongoDB. Defragments a collection and rebuilds the indexes. Internal command to ush connection pool. Converts a non-capped collection to a capped collection. Copies a database from a remote host to the current host. Creates a collection and sets collection parameters. Removes the current database. Removes indexes from a collection. Removes the specied collection from the database. Returns the md5 hash for les stored using GridFS. Flushes pending writes to the storage layer and locks the database to allow backups. Retrieves conguration options. Rotates the MongoDB logs to prevent a single le from taking too much space. Rebuilds all indexes on a collection. Changes the name of an existing collection. Repairs any errors and inconsistencies with the data storage. Modies conguration options. Shuts down the mongod (page 1021) or mongos (page 1032) process. Loads documents and indexes from data storage to memory.
855
clean clean clean (page 856) is an internal command. Warning: This command obtains a write lock on the affected database and will block other operations until it has completed.
cloneCollectionAsCapped cloneCollectionAsCapped The cloneCollectionAsCapped (page 856) command creates a new capped collection from an existing, non-capped collection within the same database. The operation does not affect the original non-capped collection. The command has the following syntax:
{ cloneCollectionAsCapped: <existing collection>, toCollection: <capped collection>, size: <capp
The command copies an existing collection and creates a new capped collection with a maximum size specied by the capped size in bytes. The name of the new capped collection must be distinct and cannot be the same as that of the original existing collection. To replace the original non-capped collection with a capped collection, use the convertToCapped (page 861) command. During the cloning, the cloneCollectionAsCapped (page 856) command exhibit the following behavior: MongoDB will transverse the documents in the original collection in natural order as theyre loaded. If the capped size specied for the new collection is smaller than the size of the original uncapped collection, then MongoDB will begin overwriting earlier documents in insertion order, which is rst in, rst out (e.g FIFO). cloneCollection cloneCollection The cloneCollection (page 856) command copies a collection from a remote server to the server where you run the command. cloneCollection (page 856) does not allow you to clone a collection through a mongos (page 1032): you must connect directly to the mongod (page 1021) instance. Parameters from Specify a resolvable hostname, and optional port number of the remote server where the specied collection resides. query Optional. A query document, in the form of a document, that lters the documents in the remote collection that cloneCollection (page 856) will copy to the current database. See db.collection.find() (page 924). copyIndexes (Boolean) Optional. true by default. When set to false the indexes on the originating server are not copied with the documents in the collection. Consider the following example:
{ cloneCollection: "users.profiles", from: "mongodb.example.net:27017", query: { active: true },
This operation copies the profiles collection from the users database on the server at mongodb.example.net. The operation only copies documents that satisfy the query { active: true } and does not copy indexes. cloneCollection (page 856) copies indexes by default, but you can disable this behavior by setting { copyIndexes: false }. The query and copyIndexes arguments are optional.
856
cloneCollection (page 856) creates a collection on the current database with the same name as the origin collection. If, in the above example, the profiles collection already exists in the local database, then MongoDB appends documents in the remote collection to the destination collection. clone clone The clone (page 857) command clone a database from a remote MongoDB instance to the current host. clone (page 857) copies the database on the remote instance with the same name as the current database. The command takes the following form:
{ clone: "db1.example.net:27017" }
Replace db1.example.net:27017 above with the resolvable hostname for the MongoDB instance you wish to copy from. Note the following behaviors: clone (page 857) can run against a slave or a non-primary member of a replica set. clone (page 857) does not snapshot the database. If any clients update the database youre copying at any point during the clone operation, the resulting database may be inconsistent. You must run clone (page 857) on the destination server. The destination server is not locked for the duration of the clone (page 857) operation. This means that clone (page 857) will occasionally yield to allow other operations to complete. See copydb (page 862) for similar functionality. Warning: This command obtains an intermittent write-lock on the destination server, that can block other operations until it completes.
closeAllDatabases closeAllDatabases closeAllDatabases (page 857) is an internal command that invalidates all cursors and closes the open database les. The next operation that uses the database will reopen the le. Warning: This command obtains a global write lock and will block other operations until it has completed.
collMod collMod New in version 2.2. collMod (page 857) makes it possible to add ags to a collection to modify the behavior of MongoDB. In the current release the only available ag is usePowerOf2Sizes (page 857). The command takes the following prototype form:
db.runCommand( {"collMod" : <collection> , "<flag>" : <value> } )
In this command substitute <collection> with the name of a collection in the current database, and <flag> and <value> with the ag and value you want to set. Use the userFlags (page 874) eld in the in db.collection.stats() (page 947) output to check enabled collection ags. usePowerOf2Sizes The usePowerOf2Sizes (page 857) ag changes the method that MongoDB uses to allocate space
857
on disk for documents in this collection. By setting usePowerOf2Sizes (page 857), you ensure that MongoDB will allocate space for documents in sizes that are powers of 2 (e.g. 4, 8, 16, 32, 64, 128, 256, 512...8388608). With usePowerOf2Sizes (page 857) MongoDB will be able to more effectively reuse space. Note: With usePowerOf2Sizes (page 857) MongoDB, allocates records that have power of 2 sizes, until record sizes equal 4 megabytes. For records larger than 4 megabytes with usePowerOf2Sizes (page 857) set, mongod (page 1021) will allocate records in full megabytes by rounding up to the nearest megabyte. usePowerOf2Sizes (page 857) is useful for collections where you will be inserting and deleting large numbers of documents to ensure that MongoDB will effectively use space on disk. Example To enable usePowerOf2Sizes (page 857) on the collection named products, use the following operation:
db.runCommand( {collMod: "products", usePowerOf2Sizes : true })
To disable usePowerOf2Sizes (page 857) on the collection products, use the following operation:
db.runCommand( { collMod: "products", "usePowerOf2Sizes": false })
Warning: Changed in version 2.2.1: usePowerOf2Sizes (page 857) now supports documents larger than 8 megabytes. If you enable usePowerOf2Sizes (page 857) you must use at least version 2.2.1. usePowerOf2Sizes (page 857) only affects subsequent allocations caused by document insertion or record relocation as a result of document growth, and does not affect existing allocations. index The index (page 858) ag changes the expiration time of a TTL Collection (page 577). Specify the key and new expiration time with a document of the form:
{keyPattern: <index_spec>, expireAfterSeconds: <seconds> }
where <index_spec> is an existing index in the collection and seconds is the number of seconds to subtract from the current time. Example To update the expiration value for a collection named sessions indexed on a lastAccess eld from 30 minutes to 60 minutes, use the following operation:
db.runCommand({collMod: "sessions", index: {keyPattern: {lastAccess:1}, expireAfterSeconds: 3600}})
Which will return the document:

{ "expireAfterSeconds_old" : 1800, "expireAfterSeconds_new" : 3600, "ok" : 1 }
On success collMod (page 857) returns a document with elds expireAfterSeconds_old and expireAfterSeconds_new set to their respective values.
858
On failure, collMod (page 857) returns a document with no expireAfterSeconds field to update if there is no existing expireAfterSeconds eld or cannot find index { **key**: 1.0 } for ns **namespace** if the specied keyPattern does not exist.
compact compact New in version 2.0. The compact (page 859) command rewrites and defragments a single collection. Additionally, the command drops all indexes at the beginning of compaction and rebuilds the indexes at the end. compact (page 859) is conceptually similar to repairDatabase (page 868), but works on a single collection rather than an entire database. The command has the following syntax:
{ compact: <collection name> }
You may also specify the following options: Parameters force (boolean) Changed in version 2.2: compact (page 859) blocks activities only for the database it is compacting. The force species whether the compact (page 859) command can run on the primary node in a replica set. Set to true to run the compact (page 859) command on the primary node in a replica set. Otherwise, the compact (page 859) command returns an error when invoked on a replica set primary because the command blocks all other activity. paddingFactor (number) New in version 2.2. Default: 1.0 Minimum: 1.0 (no padding.) Maximum: 4.0 The paddingFactor describes the record size allocated for each document as a factor of the document size, for all records compacted during the compact (page 859) operation. paddingFactor does not affect the padding of subsequent record allocations after compact (page 859) completes. If your updates increase the size of the documents, padding will increase the amount of space allocated to each document and avoid expensive document relocation operations within the data les. You can calculate the padding size by subtracting the document size from the record size or, in terms of the paddingFactor, by subtracting 1 from the paddingFactor:
padding size = (paddingFactor - 1) * <document size>.
For example, a paddingFactor of 1.0 species a padding size of 0 whereas a paddingFactor of 1.2 species a padding size of 0.2 or 20 percent (20%) of the document size. With the following command, you can use the paddingFactor option of the compact (page 859) command to set the record size to 1.1 of the document size, or a padding factor of 10 percent (10%):
859
db.runCommand ( { compact: <collection>, paddingFactor: 1.1 } )
compact (page 859) compacts existing documents, but does not reset paddingFactor statistics for the collection. After the compact (page 859) MongoDB will use the existing paddingFactor when allocating new records for documents in this collection. paddingBytes (integer) New in version 2.2. The paddingBytes sets the padding as an absolute number of bytes, for all records compacted during the compact (page 859) operation. After running compact (page 859), paddingBytes does not affect the padding of subsequent record allocations. Specifying paddingBytes can be useful if your documents start small but then increase in size signicantly. For example, if your documents are initially 40 bytes long and you grow them by 1KB, using paddingBytes: 1024 might be reasonable since using paddingFactor: 4.0 would specify a record size of 160 bytes (4.0 times the initial document size), which would only provide a padding of 120 bytes (i.e. record size of 160 bytes minus the document size). With the following command, you can use the paddingBytes option of the compact (page 859) command to set the padding size to 100 bytes on the collection named by <collection>:
db.runCommand ( { compact: <collection>, paddingBytes: 100 } )
Warning: Always have an up-to-date backup before performing server maintenance such as the compact (page 859) operation. Note the following behaviors: compact (page 859) blocks all other activity. In MongoDB 2.2, compact (page 859) blocks activities only for its database. You may view the intermediate progress either by viewing the mongod (page 1021) log le, or by running the db.currentOp() (page 971) in another shell instance. compact (page 859) compacts existing documents in the collection. However, unlike repairDatabase (page 868), compact (page 859) does not reset paddingFactor statistics for the collection. MongoDB will use the existing paddingFactor when allocating new records for documents in this collection. compact (page 859) generally uses less disk space than repairDatabase (page 868) and is faster. However,the compact (page 859) command is still slow and blocks other database use. Only use compact (page 859) during scheduled maintenance periods. If you terminate the operation with the db.killOp() (page 981) method or restart the server before the compact (page 859) operation has nished: If you have journaling enabled, the data remains consistent and usable, regardless of the state of the compact (page 859) operation. You may have to manually rebuild the indexes. If you do not have journaling enabled and the mongod (page 1021) or compact (page 859) terminates during the operation, it is impossible to guarantee that the data is in a consistent state. In either case, much of the existing free space in the collection may become un-reusable. In this scenario, you should rerun the compaction to completion to restore the use of this free space. compact (page 859) may increase the total size and number of your data les, especially when run for the rst time. However, this will not increase the total collection storage space since storage size is the amount of data allocated within the database les, and not the size/number of the les on the le system.
860
compact (page 859) requires a small amount of additional disk space while running but unlike repairDatabase (page 868) it does not free space on the le system. You may also wish to run the collStats (page 873) command before and after compaction to see how the storage space changes for the collection. compact (page 859) commands do not replicate to secondaries in a replica set: Compact each member separately. Ideally run compact (page 859) on a secondary. See option force:true above for information regarding compacting the primary. Warning: If you run compact (page 859) on a secondary, the secondary will enter a RECOVERING state to prevent clients from sending read operations during compaction. Once the operation nishes the secondary will automatically return to SECONDARY state. See state (page 841) for more information about replica set member states. Refer to the partial script for automating step down and compaction for an example of this procedure. compact (page 859) is a command issued to a mongod (page 1021). In a sharded environment, run compact (page 859) on each shard separately as a maintenance operation. Important: You cannot issue compact (page 859) against a mongos (page 1032) instance. It is not possible to compact capped collections because they dont have padding, and documents cannot grow in these collections. However, the documents of a capped collection are not subject to fragmentation. See also: repairDatabase (page 868) connPoolSync connPoolSync connPoolSync (page 861) is an internal command. convertToCapped convertToCapped The convertToCapped (page 861) command converts an existing, non-capped collection to a capped collection within the same database. The command has the following syntax:
{convertToCapped: <collection>, size: <capped size> }
convertToCapped (page 861) takes an existing collection (<collection>) and transforms it into a capped collection with a maximum size in bytes, specied to the size argument (<capped size>). During the conversion process, the convertToCapped (page 861) command exhibit the following behavior: MongoDB transverses the documents in the original collection in natural order and loads the documents into a new capped collection. If the capped size specied for the capped collection is smaller than the size of the original uncapped collection, then MongoDB will overwrite documents in the capped collection based on insertion order, or rst in, rst out order. Internally, to convert the collection, MongoDB uses the following procedure
861
cloneCollectionAsCapped (page 856) command creates the capped collection and imports the data. MongoDB drops the original collection. renameCollection (page 867) renames the new capped collection to the name of the original collection. Note: MongoDB does not support the convertToCapped (page 861) command in a sharded cluster. Warning: The convertToCapped (page 861) will not recreate indexes from the original collection on the new collection, other than the index on the _id eld. If you need indexes on this collection you will need to create these indexes after the conversion is complete. See also: create (page 863) Warning: This command obtains a global write lock and will block other operations until it has completed.
copydb copydb The copydb (page 862) command copies a database from a remote host to the current host. The command has the following syntax:
{ copydb: 1: fromhost: <hostname>, fromdb: <db>, todb: <db>, slaveOk: <bool>, username: <username>, nonce: <nonce>, key: <key> }
All of the following arguments are optional: slaveOk username nonce key You can omit the fromhost argument, to copy one database to another database within a single MongoDB instance. You must run this command on the destination, or the todb server. Be aware of the following behaviors: copydb (page 862) can run against a slave or a non-primary member of a replica set. In this case, you must set the slaveOk option to true. copydb (page 862) does not snapshot the database. If the state of the database changes at any point during the operation, the resulting database may be inconsistent. You must run copydb (page 862) on the destination server.
862
The destination server is not locked for the duration of the copydb (page 862) operation. This means that copydb (page 862) will occasionally yield to allow other operations to complete. If the remote server has authentication enabled, then you must include a username, nonce, and a key. The nonce is a one-time password that you request from the remote server using the copydbgetnonce command. The key is a hash generated as follows:
hex_md5(nonce + username + hex_md5(username + ":mongo:" + pass))
If you need to copy a database and authenticate, its easiest to use the shell helper:
db.copyDatabase(<remote_db_name>, <local_db_name>, <from_host_name>, <username>, <password>)
create create The create command explicitly creates a collection. The command uses the following syntax:
{ create: <collection_name> }
To create a capped collection limited to 40 KB, issue command in the following form:
{ create: "collection", capped: true, size: 40 * 1024 }
The options for creating capped collections are: Options capped Specify true to create a capped collection. autoIndexId Specify false to disable the automatic index created on the _id eld. Before 2.2, the default value for autoIndexId was false. See _id Fields and Indexes on Capped Collections (page 1162) for more information. size The maximum size for the capped collection. Once a capped collection reaches its max size, MongoDB will drop old documents from the database to make way for the new documents. You must specify a size argument for all capped collections. max The maximum number of documents to preserve in the capped collection. This limit is subject to the overall size of the capped collection. If a capped collection reaches its maximum size before it contains the maximum number of documents, the database will remove old documents. Thus, if you use this option, ensure that the total size for the capped collection is sufcient to contain the max. The db.createCollection() (page 970) provides a wrapper function that provides access to this functionality. Note: This command obtains a write lock on the affected database and will block other operations until it has completed. The write lock for this operation is typically short lived; however, allocations for large capped collections may take longer.
dropDatabase dropDatabase The dropDatabase (page 863) command drops a database, deleting the associated data les. dropDatabase (page 863) operates on the current database. In the shell issue the use <database> command, replacing <database> with the name of the database you wish to delete. Then use the following command form:
863
{ dropDatabase: 1 }
The mongo (page 1036) shell also provides the following equivalent helper method:
db.dropDatabase();
Warning: This command obtains a global write lock and will block other operations until it has completed.
dropIndexes dropIndexes The dropIndexes (page 864) command drops one or all indexes from the current collection. To drop all indexes, issue the command like so:
{ dropIndexes: "collection", index: "*" }
To drop a single, issue the command by specifying the name of the index you want to drop. For example, to drop the index named age_1, use the following command:
{ dropIndexes: "collection", index: "age_1" }
The shell provides a useful command helper. Heres the equivalent command:
db.collection.dropIndex("age_1");
Warning: This command obtains a write lock on the affected database and will block other operations until it has completed.
drop drop The drop (page 864) command removes an entire collection from a database. The command has following syntax:
{ drop: <collection_name> }
The mongo (page 1036) shell provides the equivalent helper method:
db.collection.drop();
Note that this command also removes any indexes associated with the dropped collection. Warning: This command obtains a write lock on the affected database and will block other operations until it has completed.
lemd5 filemd5 The filemd5 (page 864) command returns the md5 hashes for a single les stored using the GridFS specication. Client libraries use this command to verify that les are correctly written to MongoDB. The command takes the files_id of the le in question and the name of the GridFS root collection as arguments. For example:
{ filemd5: ObjectId("4f1f10e37671b50e4ecd2776"), root: "fs" }
864
fsync fsync The fsync (page 865) command forces the mongod (page 1021) process to ush all pending writes to the storage layer. mongod (page 1021) is always writing data to the storage layer as applications write more data to the database. MongoDB guarantees that it will write all data to disk within the syncdelay (page 1086) interval, which is 60 seconds by default.
{ fsync: 1 }
The fsync (page 865) operation is synchronous by default, to run fsync (page 865) asynchronously, use the following form:
{ fsync: 1, async: true }
The connection will return immediately. You can check the output of db.currentOp() (page 971) for the status of the fsync (page 865) operation. The primary use of fsync (page 865) is to lock the database during backup operations. This will ush all data to the data storage layer and block all write operations until you unlock the database. Consider the following command form:
{ fsync: 1, lock: true }
Note: You may continue to perform read operations on a database that has a fsync (page 865) lock. However, following the rst write operation all subsequent read operations wait until you unlock the database. To check on the current state of the fsync lock, use db.currentOp() (page 971). Use the following JavaScript function in the shell to test if the database is currently locked:
serverIsLocked = function () { var co = db.currentOp(); if (co && co.fsyncLock) { return true; } return false; }
After loading this function into your mongo (page 1036) shell session you can call it as follows:
serverIsLocked()
This function will return true if the database is currently locked and false if the database is not locked. To unlock the database, make a request for an unlock using the following command:
db.getSiblingDB("admin").$cmd.sys.unlock.findOne();
New in version 1.9.0: The db.fsyncLock() (page 977) and db.fsyncUnlock() (page 977) helpers in the shell. In the mongo (page 1036) shell, you may use the db.fsyncLock() (page 977) and db.fsyncUnlock() (page 977) wrappers for the fsync (page 865) lock and unlock process:
db.fsyncLock(); db.fsyncUnlock();
Note: fsync (page 865) lock is only possible on individual shards of a sharded cluster, not on the entire sharded cluster. To backup an entire sharded cluster, please read Sharded Cluster Backup Considerations (page 42).
865
If your mongod (page 1021) has journaling enabled, consider using another method (page 47) to back up your database. Note: The database cannot be locked with db.fsyncLock() (page 977) while proling is enabled. You must disable proling before locking the database with db.fsyncLock() (page 977). Disable proling using db.setProfilingLevel() (page 985) as follows in the mongo (page 1036) shell:
getParameter getParameter getParameter (page 866) is an administrative command for retrieving the value of options normally set on the command line. Issue commands against the admin database as follows:
{ getParameter: 1, <option>: 1 }
The values specied for getParameter and <option> do not affect the output. The command works with the following options: quiet notablescan logLevel syncdelay See also: setParameter (page 869) for more about these parameters. logRotate logRotate The logRotate (page 866) command is an administrative command that allows you to rotate the MongoDB logs to prevent a single logle from consuming too much disk space. You must issue the logRotate (page 866) command against the admin database in the form:
{ logRotate: 1 }
Note: Your mongod (page 1021) instance needs to be running with the --logpath [file] (page 1022) option. You may also rotate the logs by sending a SIGUSR1 signal to the mongod (page 1021) process. If your mongod (page 1021) has a process ID of 2200, heres how to send the signal on Linux:
kill -SIGUSR1 2200
logRotate (page 866) renames the existing log le by appending the current timestamp to the lename. The appended timestamp has the following form:
<YYYY>-<mm>-<DD>T<HH>-<MM>-<SS>
Then logRotate (page 866) creates a new log le with the same name as originally specied by the logpath (page 1080) setting to mongod (page 1021) or mongos (page 1032). Note: New in version 2.0.3: The logRotate (page 866) command is available to mongod (page 1021)
866
instances running on Windows systems with MongoDB release 2.0.3 and higher.
reIndex reIndex The reIndex (page 867) command rebuilds all indexes for a specied collection. Use the following syntax:
{ reIndex: "collection" }
Normally, MongoDB compacts indexes during routine updates. For most users, the reIndex (page 867) command is unnecessary. However, it may be worth running if the collection size has changed signicantly or if the indexes are consuming a disproportionate amount of disk space. Call reIndex (page 867) using the following form:
db.collection.reIndex();
Warning: This command obtains a write lock on the affected database and will block other operations until it has completed.
Note: For replica sets, reIndex (page 867) will not propagate from the primary to secondaries. reIndex (page 867) will only affect a single mongod (page 1021) instance.
renameCollection renameCollection The renameCollection (page 867) command is an administrative command that changes the name of an existing collection. You specify collections to renameCollection (page 867) in the form of a complete namespace, which includes the database name. To rename a collection, issue the renameCollection (page 867) command against the admin database in the form:
{ renameCollection: <source-namespace>, to: <target-namespace>[, dropTarget: <boolean> ] }
The dropTarget argument is optional. If you specify a collection to the to argument in a different database, the renameCollection (page 867) command will copy the collection to the new database and then drop the source collection. Parameters source-namespace Species the complete namespace of the collection to rename. to (string) Species the new namespace of the collection. dropTarget (boolean) Optional. If true, mongod (page 1021) will drop the target of renameCollection (page 867) prior to renaming the collection. Exception 10026 Raised if the source namespace does not exist. 10027 Raised if the target namespace exists and dropTarget is either false or unspecied. 15967 Raised if the target namespace is an invalid collection name. You can use renameCollection (page 867) in production environments; however: renameCollection (page 867) will block all database activity for the duration of the operation.
867
renameCollection (page 867) is incompatible with sharded collections. Warning: renameCollection (page 867) will fail if target is the name of an existing collection and you do not specify dropTarget: true. If the renameCollection (page 867) operation does not complete the target collection and indexes will not be usable and will require manual intervention to clean up. The shell helper db.collection.renameCollection() (page 945) provides a simpler interface to using this command within a database. The following is equivalent to the previous example:
db.source-namespace.renameCollection( "target" )
Warning: You cannot use renameCollection (page 867) with sharded collections. Warning: This command obtains a global write lock and will block other operations until it has completed.
repairDatabase repairDatabase Warning: In general, if you have an intact copy of your data, such as would exist on a very recent backup or an intact member of a replica set, do not use repairDatabase (page 868) or related options like db.repairDatabase() (page 984) in the mongo (page 1036) shell or mongod --repair (page 1026). Restore from an intact copy of your data.
Note: When using journaling, there is almost never any need to run repairDatabase (page 868). In the event of an unclean shutdown, the server will be able restore the data les to a pristine state automatically. The repairDatabase (page 868) command checks and repairs errors and inconsistencies with the data storage. The command is analogous to a fsck command for le systems. If your mongod (page 1021) instance is not running with journaling the system experiences an unexpected system restart or crash, and you have no other intact replica set members with this data, you should run the repairDatabase (page 868) command to ensure that there are no errors in the data storage. As a side effect, the repairDatabase (page 868) command will compact the database, as the compact (page 859) command, and also reduces the total size of the data les on disk. The repairDatabase (page 868) command will also recreate all indexes in the database. Use the following syntax:
{ repairDatabase: 1 }
Be aware that this command can take a long time to run if your database is large. In addition, it requires a quantity of free disk space equal to the size of your database. If you lack sufcient free space on the same volume, you can mount a separate volume and use that for the repair. In this case, you must run the command line and use the --repairpath (page 1027) switch to specify the folder in which to store the temporary repair les. Warning: This command obtains a global write lock and will block other operations until it has completed. This command is accessible via a number of different avenues. You may: Use the shell to run the above command, as above.
868
Use the db.repairDatabase() (page 984) in the mongo (page 1036) shell. Run mongod (page 1021) directly from your systems shell. Make sure that mongod (page 1021) isnt already running, and that you issue this command as a user that has access to MongoDBs data les. Run as:
$ mongod --repair
To add a repair path:

$ mongod --repair --repairpath /opt/vol2/data
Note: This command will fail if your database is not a master or primary. In most cases, you should recover a corrupt secondary using the data from an existing intact node. If you must repair a secondary or slave node, rst restart the node as a standalone mongod by omitting the --replSet (page 1028) or --slave (page 1029) options, as necessary.
setParameter setParameter setParameter (page 869) is an administrative command for modifying options normally set on the command line. You must issue the setParameter (page 869) command against the admin database in the form:
{ setParameter: 1, <option>: <value> }
Replace the <option> with one of the supported setParameter (page 869) options: journalCommitInterval (page 1093) logLevel (page 1094) logUserIds (page 1094) notablescan (page 1094) quiet (page 1095) replApplyBatchSize (page 1094) replIndexPrefetch (page 1094) syncdelay (page 1094) traceExceptions (page 1095) textSearchEnabled (page 1095) shutdown shutdown The shutdown (page 869) command cleans up all database resources and then terminates the process. You must issue the shutdown (page 869) command against the admin database in the form:
{ shutdown: 1 }
Note: Run the shutdown (page 869) against the admin database. When using shutdown (page 869), the connection must originate from localhost or use an authenticated connection. If the node youre trying to shut down is a replica set (page 387) primary, then the command will succeed only if there exists a secondary node whose oplog data is within 10 seconds of the primary. You can override this protection using the force option:
869
{ shutdown: 1, force: true }
Alternatively, the shutdown (page 869) command also supports a timeoutSecs argument which allows you to specify a number of seconds to wait for other members of the replica set to catch up:
{ shutdown: 1, timeoutSecs: 60 }
The equivalent mongo (page 1036) shell helper syntax looks like this:
db.shutdownServer({timeoutSecs: 60});
touch touch New in version 2.2. The touch (page 870) command loads data from the data storage layer into memory. touch (page 870) can load the data (i.e. documents,) indexes or both documents and indexes. Use this command to ensure that a collection, and/or its indexes, are in memory before another operation. By loading the collection or indexes into memory, mongod (page 1021) will ideally be able to perform subsequent operations more efciently. The touch (page 870) command has the following prototypical form:
{ touch: [collection], data: [boolean], index: [boolean] }
By default, data and index are false, and touch (page 870) will perform no operation. For example, to load both the data and the index for a collection named records, you would use the following command in the mongo (page 1036) shell:
db.runCommand({ touch: "records", data: true, index: true })
touch (page 870) will not block read and write operations on a mongod (page 1021), and can run on secondary members of replica sets. Note: Using touch (page 870) to control or tweak what a mongod (page 1021) stores in memory may displace other records data in memory and hinder performance. Use with caution in production systems. Warning: If you run touch (page 870) on a secondary, the secondary will enter a RECOVERING state to prevent clients from sending read operations during the touch (page 870) operation. When touch (page 870) nishes the secondary will automatically return to SECONDARY state. See state (page 841) for more information on replica set member states.
870
Diagnostic Commands
Diagnostic Commands
Name Description availableQueryOptions Internal command that reports on the capabilities of the current MongoDB (page 871) instance. buildInfo (page 871) Displays statistics about the MongoDB build. collStats (page 873) Reports storage utilization statics for a specied collection. connPoolStats Reports statistics on the outgoing connections from this MongoDB instance to (page 875) other MongoDB instances in the deployment. cursorInfo (page 877) Reports statistics on active cursors. dataSize (page 877) Returns the data size for a range of data. For internal use. dbHash (page 877) Internal command to support sharding. dbStats (page 877) Reports storage utilization statistics for the specied database. diagLogging (page 879) Provides a diagnostic logging. For internal use. driverOIDTest Internal command that converts an ObjectID to a string to support tests. (page 879) features (page 879) Reports on features available in the current MongoDB instance. getCmdLineOpts Returns a document with the run-time arguments to the MongoDB instance and (page 879) their parsed options. getLog (page 880) Returns recent log messages. hostInfo (page 880) Returns data that reects the underlying host system. indexStats (page 883) Experimental command that collects and aggregates statistics on all indexes. isSelf Internal command to support testing. listCommands Lists all database commands provided by the current mongod (page 1021) (page 888) instance. listDatabases Returns a document that lists all databases and returns basic database statistics. (page 888) netstat (page 888) Internal command that reports on intra-deployment connectivity. Only available for mongos (page 1032) instances. ping (page 888) Internal command that tests intra-deployment connectivity. profile (page 888) Interface for the database proler (page 1100). serverStatus Returns a collection metrics on instance-wide resource utilization and status. (page 889) top (page 906) Returns raw usage statistics for each database in the mongod (page 1021) instance. validate (page 907) Internal command that scans for a collections data and indexes for correctness. whatsmyuri (page 909) Internal command that returns information on the current client. availableQueryOptions availableQueryOptions availableQueryOptions (page 871) is an internal command that is only available on mongos (page 1032) instances. buildInfo buildInfo The buildInfo (page 871) command is an administrative command which returns a build summary for the current mongod (page 1021). buildInfo (page 871) has the following prototype form:
871
{ buildInfo: 1 }
In the mongo (page 1036) shell, call buildInfo (page 871) in the following form:
db.runCommand( { buildInfo: 1 } )
Example The output document of buildInfo (page 871) has the following form:
{ "version" : "<string>", "gitVersion" : "<string>", "sysInfo" : "<string>", "loaderFlags" : "<string>", "compilerFlags" : "<string>", "allocator" : "<string>", "versionArray" : [ <num>, <num>, <...> ], "javascriptEngine" : "<string>", "bits" : <num>, "debug" : <boolean>, "maxBsonObjectSize" : <num>, "ok" : <num> }
Consider the following documentation of the output of buildInfo (page 871): buildInfo The document returned by the buildInfo (page 871) command. buildInfo.gitVersion The commit identier that identies the state of the code used to build the mongod (page 1021). buildInfo.sysInfo A string that holds information about the operating system, hostname, kernel, date, and Boost version used to compile the mongod (page 1021). buildInfo.loaderFlags The ags passed to the loader that loads the mongod (page 1021). buildInfo.compilerFlags The ags passed to the compiler that builds the mongod (page 1021) binary. buildInfo.allocator Changed in version 2.2. The memory allocator that mongod (page 1021) uses. By default this is tcmalloc after version 2.2, and system before 2.2. buildInfo.versionArray An array that conveys version information about the mongod (page 1021) instance. See version for a more readable version of this string. buildInfo.javascriptEngine Changed in version 2.4. A string that reports the JavaScript engine used in the mongod (page 1021) instance. By default, this is V8 after version 2.4, and SpiderMonkey before 2.4. buildInfo.bits A number that reects the target processor architecture of the mongod (page 1021) binary.
872
buildInfo.debug A boolean. true when built with debugging options. buildInfo.maxBsonObjectSize A number that reports the Maximum BSON Document Size (page 1105). collStats Denition collStats The collStats (page 873) command returns a variety of storage statistics for a given collection. Use the following syntax:
{ collStats: "collection" , scale : 1024 }
Specify the collection you want statistics for, and use the scale argument to scale the output. The above example will display values in kilobytes. Examine the following example output, which uses the db.collection.stats() (page 947) helper in the mongo (page 1036) shell.
> db.users.stats() { "ns" : "app.users", "count" : 9, "size" : 432, "avgObjSize" : 48, "storageSize" : 3840, "numExtents" : 1, "nindexes" : 2, "lastExtentSize" : 3840, "paddingFactor" : 1, "flags" : 1, "totalIndexSize" : 16384, "indexSizes" : { "_id_" : 8192, "username" : 8192 }, "ok" : 1 }
// // // // // // // // //
namespace number of documents collection size in bytes average object size in bytes (pre)allocated space for the collection number of extents (contiguously allocated chunks of d number of indexes size of the most recently created extent padding can speed up updates if documents grow
// total index size in bytes // size of specific indexes in bytes
Note: The scale factor rounds values to whole numbers. This can produce unpredictable and unexpected results in some situations.
Output collStats.ns The namespace of the current collection, which follows the format [database].[collection]. collStats.count The number of objects or documents in this collection. collStats.size The size of the data stored in this collection. This value does not include the size of any indexes associated with the collection, which the totalIndexSize (page 874) eld reports. The scale argument affects this value.
873
collStats.avgObjSize The average size of an object in the collection. The scale argument affects this value. collStats.storageSize The total amount of storage allocated to this collection for document storage. The scale argument affects this value. The storageSize (page 874) does not decrease as you remove or shrink documents. collStats.numExtents The total number of contiguously allocated data le regions. collStats.nindexes The number of indexes on the collection. All collections have at least one index on the _id eld. Changed in version 2.2: Before 2.2, capped collections did not necessarily have an index on the _id eld, and some capped collections created with pre-2.2 versions of mongod (page 1021) may not have an _id index. collStats.lastExtentSize The size of the last extent allocated. The scale argument affects this value. collStats.paddingFactor The amount of space added to the end of each document at insert time. The document padding provides a small amount of extra space on disk to allow a document to grow slightly without needing to move the document. mongod (page 1021) automatically calculates this padding factor collStats.flags Changed in version 2.2: Removed in version 2.2 and replaced with the userFlags (page 874) and systemFlags (page 874) elds. Indicates the number of ags on the current collection. In version 2.0, the only ag notes the existence of an index on the _id eld. collStats.systemFlags New in version 2.2. Reports the ags on this collection that reect internal server options. Typically this value is 1 and reects the existence of an index on the _id eld. collStats.userFlags New in version 2.2. Reports the ags on this collection set by the user. In version 2.2 the only user ag is usePowerOf2Sizes (page 857). If usePowerOf2Sizes (page 857) is enabled, userFlags (page 874) will be set to 1, otherwise userFlags (page 874) will be 0. See the collMod (page 857) command for more information on setting user ags and usePowerOf2Sizes (page 857). collStats.totalIndexSize The total size of all indexes. The scale argument affects this value. collStats.indexSizes This eld species the key and size of every existing index on the collection. The scale argument affects this value. Example The following is an example of db.collection.stats() (page 947) and collStats (page 873) output:
{ "ns" : "<database>.<collection>", "count" : <number>, "size" : <number>,
874
"avgObjSize" : <number>, "storageSize" : <number>, "numExtents" : <number>, "nindexes" : <number>, "lastExtentSize" : <number>, "paddingFactor" : <number>, "systemFlags" : <bit>, "userFlags" : <bit>, "totalIndexSize" : <number>, "indexSizes" : { "_id_" : <number>, "a_1" : <number> }, "ok" : 1 }
connPoolStats Denition connPoolStats
Note: connPoolStats (page 875) only returns meaningful results for mongos (page 1032) instances and for mongod (page 1021) instances in sharded clusters. The command connPoolStats (page 875) returns information regarding the number of open connections to the current database instance, including client connections and server-to-server connections for replication and clustering. The command takes the following form:
{ connPoolStats: 1 }
The value of the argument (i.e. 1 ) does not affect the output of the command. Note: connPoolStats (page 875) only returns meaningful results for mongos (page 1032) instances and for mongod (page 1021) instances in sharded clusters.
Output connPoolStats.hosts The sub-documents of the hosts (page 875) document report connections between the mongos (page 1032) or mongod (page 1021) instance and each component mongod (page 1021) of the sharded cluster. connPoolStats.hosts.[host].available available (page 875) reports the total number of connections that the mongos (page 1032) or mongod (page 1021) could use to connect to this mongod (page 1021). connPoolStats.hosts.[host].created created (page 875) reports the number of connections that this mongos (page 1032) or mongod (page 1021) has ever created for this host. connPoolStats.replicaSets replicaSets (page 875) is a document that contains replica set information for the sharded cluster. connPoolStats.replicaSets.shard The shard (page 875) document reports on each shard within the sharded cluster
875
connPoolStats.replicaSets.[shard].host The host (page 875) eld holds an array of document that reports on each host within the shard in the replica set. These values derive from the replica set status (page 840) values. connPoolStats.replicaSets.[shard].host[n].addr addr (page 876) reports the address for the host in the sharded cluster in the format of [hostname]:[port]. connPoolStats.replicaSets.[shard].host[n].ok ok (page 876) reports false when: the mongos (page 1032) or mongod (page 1021) cannot connect to instance. the mongos (page 1032) or mongod (page 1021) received a connection exception or error. This eld is for internal use. connPoolStats.replicaSets.[shard].host[n].ismaster ismaster (page 876) reports true if this host (page 875) is the primary member of the replica set. connPoolStats.replicaSets.[shard].host[n].hidden hidden (page 876) reports true if this host (page 875) is a hidden member of the replica set. connPoolStats.replicaSets.[shard].host[n].secondary secondary (page 876) reports true if this host (page 875) is a secondary member of the replica set. connPoolStats.replicaSets.[shard].host[n].pingTimeMillis pingTimeMillis (page 876) reports the ping time in milliseconds from the mongos (page 1032) or mongod (page 1021) to this host (page 875). connPoolStats.replicaSets.[shard].host[n].tags New in version 2.2. tags (page 876) reports the tags (page 465), if this member of the set has tags congured. connPoolStats.replicaSets.[shard].master master (page 876) reports the ordinal identier of the host in the host (page 875) array that is the primary of the replica set. connPoolStats.replicaSets.[shard].nextSlave Deprecated since version 2.2. nextSlave (page 876) reports the secondary member that the mongos (page 1032) will use to service the next request for this replica set. connPoolStats.createdByType createdByType (page 876) document reports the number of each type of connection that mongos (page 1032) or mongod (page 1021) has created in all connection pools. mongos (page 1032) connect to mongod (page 1021) instances using one of three types of connections. The following sub-document reports the total number of connections by type. connPoolStats.createdByType.master master (page 876) reports the total number of connections to the primary member in each cluster. connPoolStats.createdByType.set set (page 876) reports the total number of connections to a replica set member. connPoolStats.createdByType.sync sync (page 876) reports the total number of cong database connections.
876
connPoolStats.totalAvailable totalAvailable (page 876) reports the running total of connections from the mongos (page 1032) or mongod (page 1021) to all mongod (page 1021) instances in the sharded cluster available for use. connPoolStats.totalCreated totalCreated (page 877) reports the total number of connections ever created from the mongos (page 1032) or mongod (page 1021) to all mongod (page 1021) instances in the sharded cluster. connPoolStats.numDBClientConnection numDBClientConnection (page 877) reports the total number of connections from the mongos (page 1032) or mongod (page 1021) to all of the mongod (page 1021) instances in the sharded cluster. connPoolStats.numAScopedConnection numAScopedConnection (page 877) reports the number of exception safe connections created from mongos (page 1032) or mongod (page 1021) to all mongod (page 1021) in the sharded cluster. The mongos (page 1032) or mongod (page 1021) releases these connections after receiving a socket exception from the mongod (page 1021). cursorInfo cursorInfo The cursorInfo (page 877) command returns information about current cursor allotment and use. Use the following form:
{ cursorInfo: 1 }
The value (e.g. 1 above,) does not affect the output of the command. cursorInfo (page 877) returns the total number of open cursors (totalOpen,) the size of client cursors in current use (clientCursors_size,) and the number of timed out cursors since the last server restart (timedOut.) dataSize dataSize For internal use. The dataSize (page 877) command returns the size data size for a set of data within a certain range:
{ dataSize: "database.collection", keyPattern: { field: 1 }, min: { field: 10 }, max: { field: 1
This will return a document that contains the size of all matching documents. Replace database.collection value with database and collection from your deployment. The keyPattern, min, and max parameters are options. The amount of time required to return dataSize (page 877) depends on the amount of data in the collection. dbHash dbHash dbHash (page 877) is an internal command. dbStats Denition dbStats The dbStats (page 877) command returns storage statistics for a given database. The command takes the following syntax:
877
{ dbStats: 1, scale: 1 }
The values of the options above do not affect the output of the command. The scale option allows you to specify how to scale byte values. For example, a scale value of 1024 will display the results in kilobytes rather than in bytes:
{ dbStats: 1, scale: 1024 }
Note: Because scaling rounds values to whole numbers, scaling may return unlikely or unexpected results. The time required to run the command depends on the total size of the database. Because the command must touch all data les, the command may take several seconds to run. In the mongo (page 1036) shell, the db.stats() (page 986) function provides a wrapper around dbStats (page 877). Output dbStats.db Contains the name of the database. dbStats.collections Contains a count of the number of collections in that database. dbStats.objects Contains a count of the number of objects (i.e. documents) in the database across all collections. dbStats.avgObjSize The average size of each document in bytes. This is the dataSize (page 878) divided by the number of documents. dbStats.dataSize The total size of the data held in this database including the padding factor. The scale argument affects this value. The dataSize (page 878) will not decrease when documents shrink, but will decrease when you remove documents. dbStats.storageSize The total amount of space allocated to collections in this database for document storage. The scale argument affects this value. The storageSize (page 878) does not decrease as you remove or shrink documents. dbStats.numExtents Contains a count of the number of extents in the database across all collections. dbStats.indexes Contains a count of the total number of indexes across all collections in the database. dbStats.indexSize The total size of all indexes created on this database. The scale arguments affects this value. dbStats.fileSize The total size of the data les that hold the database. This value includes preallocated space and the padding factor. The value of fileSize (page 878) only reects the size of the data les for the database and not the namespace le. The scale argument affects this value. dbStats.nsSizeMB The total size of the namespace les (i.e. that end with .ns) for this database. You cannot change the size of the namespace le after creating a database, but you can change the default size for all new namespace les with the nssize (page 1084) runtime option.
878
See also: The nssize (page 1084) option, and Maximum Namespace File Size (page 1105) dbStats.dataFileVersion New in version 2.4. Document that contains information about the on-disk format of the data les for the database. dbStats.dataFileVersion.major New in version 2.4. The major version number for the on-disk format of the data les for the database. dbStats.dataFileVersion.minor New in version 2.4. The minor version number for the on-disk format of the data les for the database. diagLogging diagLogging diagLogging (page 879) is an internal command. Warning: This command obtains a write lock on the affected database and will block other operations until it has completed.
driverOIDTest driverOIDTest driverOIDTest (page 879) is an internal command. features features features (page 879) is an internal command that returns the build-level feature settings. getCmdLineOpts getCmdLineOpts The getCmdLineOpts (page 879) command returns a document containing command line options used to start the given mongod (page 1021):
{ getCmdLineOpts: 1 }
This command returns a document with two elds, argv and parsed. The argv eld contains an array with each item from the command string used to invoke mongod (page 1021). The document in the parsed eld includes all runtime options, including those parsed from the command line and those specied in the conguration le, if specied. Consider the following example output of getCmdLineOpts (page 879):
{ "argv" : [ "/usr/bin/mongod", "--config", "/etc/mongodb.conf", "--fork" ], "parsed" : {
879
"bind_ip" : "127.0.0.1", "config" : "/etc/mongodb/mongodb.conf", "dbpath" : "/srv/mongodb", "fork" : true, "logappend" : "true", "logpath" : "/var/log/mongodb/mongod.log", "quiet" : "true" }, "ok" : 1 }
http://docs.mongodb.org/manual/administration/import-export/ getLog getLog The getLog (page 880) command returns a document with a log array that contains recent messages from the mongod (page 1021) process log. The getLog (page 880) command has the following syntax:
{ getLog: <log> }
Replace <log> with one of the following values: global - returns the combined output of all recent log entries. rs - if the mongod (page 1021) is part of a replica set, getLog (page 880) will return recent notices related to replica set activity. startupWarnings - will return logs that may contain errors or warnings from MongoDBs log from when the current process started. If mongod (page 1021) started without warnings, this lter may return an empty array. You may also specify an asterisk (e.g. *) as the <log> value to return a list of available log lters. The following interaction from the mongo (page 1036) shell connected to a replica set:
db.adminCommand({getLog: "*" }) { "names" : [ "global", "rs", "startupWarnings" ], "ok" : 1 }
getLog (page 880) returns events from a RAM cache of the mongod (page 1021) events and does not read log data from the log le. hostInfo hostInfo New in version 2.2. Returns A document with information about the underlying system that the mongod (page 1021) or mongos (page 1032) runs on. Some of the returned elds are only included on some platforms. You must run the hostInfo (page 880) command, which takes no arguments, against the admin database. Consider the following invocations of hostInfo (page 880):
db.hostInfo() db.adminCommand( { "hostInfo" : 1 } )
In the mongo (page 1036) shell you can use db.hostInfo() (page 980) as a helper to access hostInfo (page 880). The output of hostInfo (page 880) on a Linux system will resemble the following:
{ "system" : { "currentTime" : ISODate("<timestamp>"),
880
"hostname" : "<hostname>", "cpuAddrSize" : <number>, "memSizeMB" : <number>, "numCores" : <number>, "cpuArch" : "<identifier>", "numaEnabled" : <boolean> }, "os" : { "type" : "<string>", "name" : "<string>", "version" : "<string>" }, "extra" : { "versionString" : "<string>", "libcVersion" : "<string>", "kernelVersion" : "<string>", "cpuFrequencyMHz" : "<string>", "cpuFeatures" : "<string>", "pageSize" : <number>, "numPages" : <number>, "maxOpenFiles" : <number> }, "ok" : <return> }
Consider the following documentation of these elds: hostInfo The document returned by the hostInfo (page 880). hostInfo.system A sub-document about the underlying environment of the system running the mongod (page 1021) or mongos (page 1032) hostInfo.system.currentTime A time stamp of the current system time. hostInfo.system.hostname The system name, which should correspond to the output of hostname -f on Linux systems. hostInfo.system.cpuAddrSize A number reecting the architecture of the system. Either 32 or 64. hostInfo.system.memSizeMB The total amount of system memory (RAM) in megabytes. hostInfo.system.numCores The total number of available logical processor cores. hostInfo.system.cpuArch A string that represents the system architecture. Either x86 or x86_64. hostInfo.system.numaEnabled A boolean value. false if NUMA is interleaved (i.e. disabled,) otherwise true. hostInfo.os A sub-document that contains information about the operating system running the mongod (page 1021) and mongos (page 1032). hostInfo.os.type A string representing the type of operating system, such as Linux or Windows.
881
hostInfo.os.name If available, returns a display name for the operating system. hostInfo.os.version If available, returns the name of the distribution or operating system. hostInfo.extra A sub-document with extra information about the operating system and the underlying hardware. The content of the extra (page 882) sub-document depends on the operating system. hostInfo.extra.versionString A complete string of the operating system version and identication. On Linux and OS X systems, this contains output similar to uname -a. hostInfo.extra.libcVersion The release of the system libc. libcVersion (page 882) only appears on Linux systems. hostInfo.extra.kernelVersion The release of the Linux kernel in current use. kernelVersion (page 882) only appears on Linux systems. hostInfo.extra.alwaysFullSync alwaysFullSync (page 882) only appears on OS X systems. hostInfo.extra.nfsAsync nfsAsync (page 882) only appears on OS X systems. hostInfo.extra.cpuFrequencyMHz Reports the clock speed of the systems processor in megahertz. hostInfo.extra.cpuFeatures Reports the processor feature ags. On Linux systems this the same information that http://docs.mongodb.org/manual/proc/cpuinfo includes in the flags elds. hostInfo.extra.pageSize Reports the default system page size in bytes. hostInfo.extra.numPages numPages (page 882) only appears on Linux systems. hostInfo.extra.maxOpenFiles Reports the current system limits on open le handles. See Linux ulimit Settings (page 109) for more information. maxOpenFiles (page 882) only appears on Linux systems. hostInfo.extra.scheduler Reports the active I/O scheduler. scheduler (page 882) only appears on OS X systems.
indexStats
Denition (page 883) Output (page 883) Example (page 884) Additional Resources (page 887)
882
Denition indexStats The indexStats (page 883) command aggregates statistics for the B-tree data structure that stores data for a MongoDB index. Warning: This command is not intended for production deployments. The command can be run only on a mongod --enableExperimentalIndexStatsCmd option. To aggregate statistics, issue the command like so:
db.runCommand( { indexStats: "<collection>", index: "<index name>" } )
(page
1021)
instance
that
uses
the
Output The db.collection.indexStats() method and equivalent indexStats (page 883) command aggregate statistics for the B-tree data structure that stores data for a MongoDB index. The commands aggregate statistics rstly for the entire B-tree and secondly for each individual level of the B-tree. The output displays the following values. indexStats.index The index name (page 335). indexStats.version The index version. For more information on index version numbers, see the v option in db.collection.ensureIndex() (page 921). indexStats.isIdIndex If true, the index is the default _id index for the collection. indexStats.keyPattern The indexed keys. indexStats.storageNs The namespace of the indexs underlying storage. indexStats.bucketBodyBytes The xed size, in bytes, of a B-tree bucket in the index, not including the record header. All indexes for a given version have the same value for this eld. MongoDB allocates xed size buckets on disk. indexStats.depth The number of levels in the B-tree, not including the root level. indexStats.overall This section of the output displays statistics for the entire B-tree. indexStats.overall.numBuckets The number of buckets in the entire B-tree, including all levels. indexStats.overall.keyCount Statistics about the number of keys in a bucket, evaluated on a per-bucket level. indexStats.overall.usedKeyCount Statistics about the number of used keys in a bucket, evaluated on a per-bucket level. Used keys are keys not marked as deleted. indexStats.overall.bsonRatio Statistics about the percentage of the bucket body that is occupied by the key objects themselves, excluding associated metadata.
883
For example, if you have the document { name: }, the key object is the string Bob Smith.
"Bob Smith" } and an index on { name:
indexStats.overall.keyNodeRatio Statistics about the percentage of the bucket body that is occupied by the key node objects (the metadata and links pertaining to the keys). This does not include the key itself. In the current implementation, a key nodes objects consist of: the pointer to the key data (in the same bucket), the pointer to the record the key is for, and the pointer to a child bucket. indexStats.overall.fillRatio The sum of the bsonRatio (page 883) and the keyNodeRatio (page 884). This shows how full the buckets are. This will be much higher for indexes with sequential inserts. indexStats.perLevel This section of the output displays statistics for each level of the B-tree separately, starting with the root level. This section displays a different document for each B-tree level. indexStats.perLevel.numBuckets The number of buckets at this level of the B-tree. indexStats.perLevel.keyCount Statistics about the number of keys in a bucket, evaluated on a per-bucket level. indexStats.perLevel.usedKeyCount Statistics about the number of used keys in a bucket, evaluated on a per-bucket level. Used keys are keys not marked as deleted. indexStats.perLevel.bsonRatio Statistics about the percentage of the bucket body that is occupied by the key objects themselves, excluding associated metadata. indexStats.perLevel.keyNodeRatio Statistics about the percentage of the bucket body that is occupied by the key node objects (the metadata and links pertaining to the keys). indexStats.perLevel.fillRatio The sum of the bsonRatio (page 884) and the keyNodeRatio (page 884). This shows how full the buckets are. This will be much higher in the following cases: For indexes with sequential inserts, such as the _id index when using ObjectId keys. For indexes that were recently built in the foreground with existing data. If you recently ran compact (page 859) or --repair (page 1026). Example The following is an example of db.collection.indexStats() and indexStats (page 883) output.
{ "index" : "type_1_traits_1", "version" : 1, "isIdIndex" : false, "keyPattern" : { "type" : 1, "traits" : 1 }, "storageNs" : "test.animals.$type_1_traits_1", "bucketBodyBytes" : 8154, "depth" : 2, "overall" : {
884
"numBuckets" : 45513, "keyCount" : { "count" : NumberLong(45513), "mean" : 253.89602970579836, "stddev" : 21.784799875240708, "min" : 52, "max" : 290, "quantiles" : { "0.01" : 201.99785091648775, // ... "0.99" : 289.9999655156967 } }, "usedKeyCount" : { "count" : NumberLong(45513), // ... "quantiles" : { "0.01" : 201.99785091648775, // ... "0.99" : 289.9999655156967 } }, "bsonRatio" : { "count" : NumberLong(45513), // ... "quantiles" : { "0.01" : 0.4267797891997124, // ... "0.99" : 0.5945548174629648 } }, "keyNodeRatio" : { "count" : NumberLong(45513), // ... "quantiles" : { "0.01" : 0.3963656628236211, // ... "0.99" : 0.5690457993930765 } }, "fillRatio" : { "count" : NumberLong(45513), // ... "quantiles" : { "0.01" : 0.9909134214926929, // ... "0.99" : 0.9960755457453732 } } }, "perLevel" : [ { "numBuckets" : 1, "keyCount" : { "count" : NumberLong(1), "mean" : 180, "stddev" : 0, "min" : 180,
885
"max" : 180 }, "usedKeyCount" : { "count" : NumberLong(1), // ... "max" : 180 }, "bsonRatio" : { "count" : NumberLong(1), // ... "max" : 0.3619082658817758 }, "keyNodeRatio" : { "count" : NumberLong(1), // ... "max" : 0.35320088300220753 }, "fillRatio" : { "count" : NumberLong(1), // ... "max" : 0.7151091488839834 } }, { "numBuckets" : 180, "keyCount" : { "count" : NumberLong(180), "mean" : 250.84444444444443, "stddev" : 26.30057503009355, "min" : 52, "max" : 290 }, "usedKeyCount" : { "count" : NumberLong(180), // ... "max" : 290 }, "bsonRatio" : { "count" : NumberLong(180), // ... "max" : 0.5945548197203826 }, "keyNodeRatio" : { "count" : NumberLong(180), // ... "max" : 0.5690458670591121 }, "fillRatio" : { "count" : NumberLong(180), // ... "max" : 0.9963208241353937 } }, { "numBuckets" : 45332, "keyCount" : { "count" : NumberLong(45332), "mean" : 253.90977675813994,
886
"stddev" : 21.761620836279018, "min" : 167, "max" : 290, "quantiles" : { "0.01" : 202.0000012563603, // ... "0.99" : 289.99996486571894 } }, "usedKeyCount" : { "count" : NumberLong(45332), // ... "quantiles" : { "0.01" : 202.0000012563603, // ... "0.99" : 289.99996486571894 } }, "bsonRatio" : { "count" : NumberLong(45332), // ... "quantiles" : { "0.01" : 0.42678446958950583, // ... "0.99" : 0.5945548175411283 } }, "keyNodeRatio" : { "count" : NumberLong(45332), // ... "quantiles" : { "0.01" : 0.39636988227885306, // ... "0.99" : 0.5690457981176729 } }, "fillRatio" : { "count" : NumberLong(45332), // ... "quantiles" : { "0.01" : 0.9909246995605362, // ... "0.99" : 0.996075546919481 } } } ], "ok" : 1 }
Additional Resources For more information on the commands limits and output, see the following: The equivalent db.collection.indexStats() method, indexStats (page 882), and https://github.com/10gen-labs/storage-viz#readme.
887
isSelf _isSelf _isSelf (page 888) is an internal command. listCommands listCommands The listCommands (page 888) command generates a list of all database commands implemented for the current mongod (page 1021) instance. listDatabases listDatabases The listDatabases (page 888) command provides a list of existing databases along with basic statistics about them:
{ listDatabases: 1 }
The value (e.g. 1) does not affect the output of the command. listDatabases (page 888) returns a document for each database Each document contains a name eld with the database name, a sizeOnDisk eld with the total size of the database le on disk in bytes, and an empty eld specifying whether the database has any data. netstat netstat netstat (page 888) is an internal command that is only available on mongos (page 1032) instances. ping ping The ping (page 888) command is a no-op used to test whether a server is responding to commands. This command will return immediately even if the server is write-locked:
{ ping: 1 }
The value (e.g. 1 above,) does not impact the behavior of the command. prole profile Use the profile (page 888) command to enable, disable, or change the query proling level. This allows administrators to capture data regarding performance. The database proling system can impact performance and can allow the server to write the contents of queries to the log. Your deployment should carefully consider the security implications of this. Consider the following prototype syntax:
{ profile: <level> }
The following proling levels are available: Level -1 0 1 2 Setting No change. Returns the current prole level. Off. No proling. On. Only includes slow operations. On. Includes all operations.
You may optionally set a threshold in milliseconds for proling using the slowms option, as follows:
888
{ profile: 1, slowms: 200 }
mongod (page 1021) writes the output of the database proler to the system.profile collection. mongod (page 1021) records queries that take longer than the slowms (page 1085) to the server log even when the database proler is not active. See also: Additional documentation regarding database proling Database Proling (page 96). See also: db.getProfilingStatus() (page 979) and db.setProfilingLevel() (page 985) provide wrappers around this functionality in the mongo (page 1036) shell. Note: The database cannot be locked with db.fsyncLock() (page 977) while proling is enabled. You must disable proling before locking the database with db.fsyncLock() (page 977). Disable proling using db.setProfilingLevel() (page 985) as follows in the mongo (page 1036) shell:
Note: This command obtains a write lock on the affected database and will block other operations until it has completed. However, the write lock is only held while enabling or disabling the proler. This is typically a short operation.
serverStatus Denition serverStatus The serverStatus (page 889) command returns a document that provides an overview of the database processs state. Most monitoring applications run this command at a regular interval to collection statistics about the instance:
{ serverStatus: 1 }
The value (i.e. 1 above), does not affect the operation of the command. Changed in version 2.4: In 2.4 you can dynamically suppress portions of the serverStatus (page 889) output, or include suppressed sections by adding elds to the command document as in the following examples:
db.runCommand( { serverStatus: 1, repl: 0, indexCounters: 0 } ) db.runCommand( { serverStatus: 1, workingSet: 1, metrics: 0, locks: 0 } )
serverStatus (page 889) includes all elds by default, except workingSet (page 902), by default. Note: You may only dynamically include top-level elds from the serverStatus (page 889) document that are not included by default. You can exclude any eld that serverStatus (page 889) includes by default. See also: db.serverStatus() (page 984) and http://docs.mongodb.org/manual/reference/server-status
889
Output The serverStatus (page 889) command returns a collection of information that reects the databases status. These data are useful for diagnosing and assessing the performance of your MongoDB instance. This reference catalogs each datum included in the output of this command and provides context for using this data to more effectively administer your database. See also: Much of the output of serverStatus (page 889) is also displayed dynamically by mongostat (page 1063). See the mongostat (page 1063) command for more information.
For examples of the serverStatus (page 889) output, see http://docs.mongodb.org/manual/reference/server-st Instance Information Example output of the instance information elds. serverStatus.host The host (page 890) eld contains the systems hostname. In Unix/Linux systems, this should be the same as the output of the hostname command. serverStatus.version The version (page 890) eld contains the version of MongoDB running on the current mongod (page 1021) or mongos (page 1032) instance. serverStatus.process The process (page 890) eld identies which kind of MongoDB instance is running. Possible values are: mongos (page 1032) mongod (page 1021) serverStatus.uptime The value of the uptime (page 890) eld corresponds to the number of seconds that the mongos (page 1032) or mongod (page 1021) process has been active. serverStatus.uptimeEstimate uptimeEstimate (page 890) provides the uptime as calculated from MongoDBs internal course-grained time keeping system. serverStatus.localTime The localTime (page 890) value is the current time, according to the server, in UTC specied in an ISODate format. locks New in version 2.1.2: All locks (page 890) statuses rst appeared in the 2.1.2 development release for the 2.2 series. Example output of the locks elds. serverStatus.locks The locks (page 890) document contains sub-documents that provides a granular report on MongoDB database-level lock use. All values are of the NumberLong() type. Generally, elds named: R refer to the global read lock, W refer to the global write lock,
890
r refer to the database specic read lock, and w refer to the database specic write lock. If a document does not have any elds, it means that no locks have existed with this context since the last time the mongod (page 1021) started. serverStatus.locks.. A eld named . holds the rst document in locks (page 890) that contains information about the global lock. serverStatus.locks...timeLockedMicros The timeLockedMicros (page 891) document reports the amount of time in microseconds that a lock has existed in all databases in this mongod (page 1021) instance. serverStatus.locks...timeLockedMicros.R The R eld reports the amount of time in microseconds that any database has held the global read lock. serverStatus.locks...timeLockedMicros.W The W eld reports the amount of time in microseconds that any database has held the global write lock. serverStatus.locks...timeLockedMicros.r The r eld reports the amount of time in microseconds that any database has held the local read lock. serverStatus.locks...timeLockedMicros.w The w eld reports the amount of time in microseconds that any database has held the local write lock. serverStatus.locks...timeAcquiringMicros The timeAcquiringMicros (page 891) document reports the amount of time in microseconds that operations have spent waiting to acquire a lock in all databases in this mongod (page 1021) instance. serverStatus.locks...timeAcquiringMicros.R The R eld reports the amount of time in microseconds that any database has spent waiting for the global read lock. serverStatus.locks...timeAcquiringMicros.W The W eld reports the amount of time in microseconds that any database has spent waiting for the global write lock. serverStatus.locks.admin The admin (page 891) document contains two sub-documents that report data regarding lock use in the admin database. serverStatus.locks.admin.timeLockedMicros The timeLockedMicros (page 891) document reports the amount of time in microseconds that locks have existed in the context of the admin database. serverStatus.locks.admin.timeLockedMicros.r The r eld reports the amount of time in microseconds that the admin database has held the read lock. serverStatus.locks.admin.timeLockedMicros.w The w eld reports the amount of time in microseconds that the admin database has held the write lock. serverStatus.locks.admin.timeAcquiringMicros The timeAcquiringMicros (page 891) document reports on the amount of eld time in microseconds that operations have spent waiting to acquire a lock for the admin database. serverStatus.locks.admin.timeAcquiringMicros.r The r eld reports the amount of time in microseconds that operations have spent waiting to acquire a read lock on the admin database. serverStatus.locks.admin.timeAcquiringMicros.w The w eld reports the amount of time in microseconds that operations have spent waiting to acquire a write lock on the admin database.
891
serverStatus.locks.local The local (page 891) document contains two sub-documents that report data regarding lock use in the local database. The local database contains a number of instance specic data, including the oplog for replication. serverStatus.locks.local.timeLockedMicros The timeLockedMicros (page 892) document reports on the amount of time in microseconds that locks have existed in the context of the local database. serverStatus.locks.local.timeLockedMicros.r The r eld reports the amount of time in microseconds that the local database has held the read lock. serverStatus.locks.local.timeLockedMicros.w The w eld reports the amount of time in microseconds that the local database has held the write lock. serverStatus.locks.local.timeAcquiringMicros The timeAcquiringMicros (page 892) document reports on the amount of time in microseconds that operations have spent waiting to acquire a lock for the local database. serverStatus.locks.local.timeAcquiringMicros.r The r eld reports the amount of time in microseconds that operations have spent waiting to acquire a read lock on the local database. serverStatus.locks.local.timeAcquiringMicros.w The w eld reports the amount of time in microseconds that operations have spent waiting to acquire a write lock on the local database. serverStatus.locks.<database> For each additional database locks (page 890) includes a document that reports on the lock use for this database. The names of these documents reect the database name itself. serverStatus.locks.<database>.timeLockedMicros The timeLockedMicros (page 892) document reports on the amount of time in microseconds that locks have existed in the context of the <database> database. serverStatus.locks.<database>.timeLockedMicros.r The r eld reports the amount of time in microseconds that the <database> database has held the read lock. serverStatus.locks.<database>.timeLockedMicros.w The w eld reports the amount of time in microseconds that the <database> database has held the write lock. serverStatus.locks.<database>.timeAcquiringMicros The timeAcquiringMicros (page 892) document reports on the amount of time in microseconds that operations have spent waiting to acquire a lock for the <database> database. serverStatus.locks.<database>.timeAcquiringMicros.r The r eld reports the amount of time in microseconds that operations have spent waiting to acquire a read lock on the <database> database. serverStatus.locks.<database>.timeAcquiringMicros.w The w eld reports the amount of time in microseconds that operations have spent waiting to acquire a write lock on the <database> database. globalLock Example output of the globalLock elds. serverStatus.globalLock The globalLock (page 892) data structure contains information regarding the databases current lock state, historical lock status, current operation queue, and the number of active clients.
892
serverStatus.globalLock.totalTime The value of totalTime (page 892) represents the time, in microseconds, since the database last started and creation of the globalLock (page 892). This is roughly equivalent to total server uptime. serverStatus.globalLock.lockTime The value of lockTime (page 893) represents the time, in microseconds, since the database last started, that the globalLock (page 892) has been held. Consider this value in combination with the value of totalTime (page 892). MongoDB aggregates these values in the ratio (page 893) value. If the ratio (page 893) value is small but totalTime (page 892) is high the globalLock (page 892) has typically been held frequently for shorter periods of time, which may be indicative of a more normal use pattern. If the lockTime (page 893) is higher and the totalTime (page 892) is smaller (relatively,) then fewer operations are responsible for a greater portion of servers use (relatively.) serverStatus.globalLock.ratio Changed in version 2.2: ratio (page 893) was removed. See locks (page 890). The value of ratio (page 893) displays the relationship between lockTime (page 893) and totalTime (page 892). Low values indicate that operations have held the globalLock (page 892) frequently for shorter periods of time. High values indicate that operations have held globalLock (page 892) infrequently for longer periods of time. serverStatus.globalLock.currentQueue The currentQueue (page 893) data structure value provides more granular information concerning the number of operations queued because of a lock. serverStatus.globalLock.currentQueue.total The value of total (page 893) provides a combined total of operations queued waiting for the lock. A consistently small queue, particularly of shorter operations should cause no concern. Also, consider this value in light of the size of queue waiting for the read lock (e.g. readers (page 893)) and write-lock (e.g. writers (page 893)) individually. serverStatus.globalLock.currentQueue.readers The value of readers (page 893) is the number of operations that are currently queued and waiting for the read-lock. A consistently small read-queue, particularly of shorter operations should cause no concern. serverStatus.globalLock.currentQueue.writers The value of writers (page 893) is the number of operations that are currently queued and waiting for the write-lock. A consistently small write-queue, particularly of shorter operations is no cause for concern. globalLock.activeClients serverStatus.globalLock.activeClients The activeClients (page 893) data structure provides more granular information about the number of connected clients and the operation types (e.g. read or write) performed by these clients. Use this data to provide context for the currentQueue (page 893) data. serverStatus.globalLock.activeClients.total The value of total (page 893) is the total number of active client connections to the database. This combines clients that are performing read operations (e.g. readers (page 893)) and clients that are performing write operations (e.g. writers (page 893)). serverStatus.globalLock.activeClients.readers The value of readers (page 893) contains a count of the active client connections performing read operations. serverStatus.globalLock.activeClients.writers The value of writers (page 893) contains a count of active client connections performing write operations.
893
mem Example output of the memory elds. serverStatus.mem The mem (page 894) data structure holds information regarding the target system architecture of mongod (page 1021) and current memory use. serverStatus.mem.bits The value of bits (page 894) is either 64 or 32, depending on which target architecture specied during the mongod (page 1021) compilation process. In most instances this is 64, and this value does not change over time. serverStatus.mem.resident The value of resident (page 894) is roughly equivalent to the amount of RAM, in megabytes (MB), currently used by the database process. In normal use this value tends to grow. In dedicated database servers this number tends to approach the total amount of system memory. serverStatus.mem.virtual virtual (page 894) displays the quantity, in megabytes (MB), of virtual memory used by the mongod (page 1021) process. With journaling enabled, the value of virtual (page 894) is at least twice the value of mapped (page 894). If virtual (page 894) value is signicantly larger than mapped (page 894) (e.g. 3 or more times), this may indicate a memory leak. serverStatus.mem.supported supported (page 894) is true when the underlying system supports extended memory information. If this value is false and the system does not support extended memory information, then other mem (page 894) values may not be accessible to the database server. serverStatus.mem.mapped The value of mapped (page 894) provides the amount of mapped memory, in megabytes (MB), by the database. Because MongoDB uses memory-mapped les, this value is likely to be to be roughly equivalent to the total size of your database or databases. serverStatus.mem.mappedWithJournal mappedWithJournal (page 894) provides the amount of mapped memory, in megabytes (MB), including the memory used for journaling. This value will always be twice the value of mapped (page 894). This eld is only included if journaling is enabled. connections Example output of the connections elds. serverStatus.connections The connections (page 894) sub document data regarding the current connection status and availability of the database server. Use these values to asses the current load and capacity requirements of the server. serverStatus.connections.current The value of current (page 894) corresponds to the number of connections to the database server from clients. This number includes the current shell session. Consider the value of available (page 894) to add more context to this datum. This gure will include the current shell connection as well as any inter-node connections to support a replica set or sharded cluster.
894
serverStatus.connections.available available (page 894) provides a count of the number of unused available connections that the database can provide. Consider this value in combination with the value of current (page 894) to understand the connection load on the database, and the Linux ulimit Settings (page 109) document for more information about system thresholds on available connections. serverStatus.connections.totalCreated totalCreated (page 895) provides a count of all connections created to the server. This number includes connections that have since closed. extra_info Example output of the extra_info elds. serverStatus.extra_info The extra_info (page 895) data structure holds data collected by the mongod (page 1021) instance about the underlying system. Your system may only report a subset of these elds. serverStatus.extra_info.note The eld note (page 895) reports that the data in this structure depend on the underlying platform, and has the text: elds vary by platform. serverStatus.extra_info.heap_usage_bytes The heap_usage_bytes (page 895) eld is only available on Unix/Linux systems, and reports the total size in bytes of heap space used by the database process. serverStatus.extra_info.page_faults The page_faults (page 895) eld is only available on Unix/Linux systems, and reports the total number of page faults that require disk operations. Page faults refer to operations that require the database server to access data which isnt available in active memory. The page_faults (page 895) counter may increase dramatically during moments of poor performance and may correlate with limited memory environments and larger data sets. Limited and sporadic page faults do not necessarily indicate an issue. indexCounters Example output of the indexCounters elds. serverStatus.indexCounters Changed in version 2.2: Previously, data in the indexCounters (page 895) document reported sampled data, and were only useful in relative comparison to each other, because they could not reect absolute index use. In 2.2 and later, these data reect actual index use. Changed in version 2.4: Fields previously in the btree sub-document of indexCounters (page 895) are now elds in the indexCounters (page 895) document. The indexCounters (page 895) data structure reports information regarding the state and use of indexes in MongoDB. serverStatus.indexCounters.accesses accesses (page 895) reports the number of times that operations have accessed indexes. This value is the combination of the hits (page 895) and misses (page 896). Higher values indicate that your database has indexes and that queries are taking advantage of these indexes. If this number does not grow over time, this might indicate that your indexes do not effectively support your use.
895
serverStatus.indexCounters.hits The hits (page 895) value reects the number of times that an index has been accessed and mongod (page 1021) is able to return the index from memory. A higher value indicates effective index use. hits (page 895) values that represent a greater proportion of the accesses (page 895) value, tend to indicate more effective index conguration. serverStatus.indexCounters.misses The misses (page 896) value represents the number of times that an operation attempted to access an index that was not in memory. These misses, do not indicate a failed query or operation, but rather an inefcient use of the index. Lower values in this eld indicate better index use and likely overall performance as well. serverStatus.indexCounters.resets The resets (page 896) value reects the number of times that the index counters have been reset since the database last restarted. Typically this value is 0, but use this value to provide context for the data specied by other indexCounters (page 895) values. serverStatus.indexCounters.missRatio The missRatio (page 896) value is the ratio of hits (page 895) to misses (page 896) misses. This value is typically 0 or approaching 0. backgroundFlushing Example output of the backgroundFlushing elds. serverStatus.backgroundFlushing mongod (page 1021) periodically ushes writes to disk. In the default conguration, this happens every 60 seconds. The backgroundFlushing (page 896) data structure contains data regarding these operations. Consider these values if you have concerns about write performance and journaling (page 900). serverStatus.backgroundFlushing.flushes flushes (page 896) is a counter that collects the number of times the database has ushed all writes to disk. This value will grow as database runs for longer periods of time. serverStatus.backgroundFlushing.total_ms The total_ms (page 896) value provides the total number of milliseconds (ms) that the mongod (page 1021) processes have spent writing (i.e. ushing) data to disk. Because this is an absolute value, consider the value of flushes (page 896) and average_ms (page 896) to provide better context for this datum. serverStatus.backgroundFlushing.average_ms The average_ms (page 896) value describes the relationship between the number of ushes and the total amount of time that the database has spent writing data to disk. The larger flushes (page 896) is, the more likely this value is likely to represent a normal, time; however, abnormal data can skew this value. Use the last_ms (page 896) to ensure that a high average is not skewed by transient historical issue or a random write distribution. serverStatus.backgroundFlushing.last_ms The value of the last_ms (page 896) eld is the amount of time, in milliseconds, that the last ush operation took to complete. Use this value to verify that the current performance of the server and is in line with the historical data provided by average_ms (page 896) and total_ms (page 896). serverStatus.backgroundFlushing.last_finished The last_finished (page 896) eld provides a timestamp of the last completed ush operation in the ISODate format. If this value is more than a few minutes old relative to your servers current time and accounting for differences in time zone, restarting the database may result in some data loss. Also consider ongoing operations that might skew this value by routinely block write operations.
896
cursors Example output of the cursors elds. serverStatus.cursors The cursors (page 897) data structure contains data regarding cursor state and use. serverStatus.cursors.totalOpen totalOpen (page 897) provides the number of cursors that MongoDB is maintaining for clients. Because MongoDB exhausts unused cursors, typically this value small or zero. However, if there is a queue, stale tailable cursors, or a large number of operations this value may rise. serverStatus.cursors.clientCursors_size Deprecated since version 1.x: See totalOpen (page 897) for this datum. serverStatus.cursors.timedOut timedOut (page 897) provides a counter of the total number of cursors that have timed out since the server process started. If this number is large or growing at a regular rate, this may indicate an application error. network Example output of the network elds. serverStatus.network The network (page 897) data structure contains data regarding MongoDBs network use. serverStatus.network.bytesIn The value of the bytesIn (page 897) eld reects the amount of network trafc, in bytes, received by this database. Use this value to ensure that network trafc sent to the mongod (page 1021) process is consistent with expectations and overall inter-application trafc. serverStatus.network.bytesOut The value of the bytesOut (page 897) eld reects the amount of network trafc, in bytes, sent from this database. Use this value to ensure that network trafc sent by the mongod (page 1021) process is consistent with expectations and overall inter-application trafc. serverStatus.network.numRequests The numRequests (page 897) eld is a counter of the total number of distinct requests that the server has received. Use this value to provide context for the bytesIn (page 897) and bytesOut (page 897) values to ensure that MongoDBs network utilization is consistent with expectations and application use. repl Example output of the repl elds. serverStatus.repl The repl (page 897) data structure contains status information for MongoDBs replication (i.e. replica set) conguration. These values only appear when the current host has replication enabled. See Replica Set Fundamental Concepts (page 387) for more information on replication. serverStatus.repl.setName The setName (page 897) eld contains a string with the name of the current replica set. This value reects the --replSet (page 1028) command line argument, or replSet (page 1088) value in the conguration le. See Replica Set Fundamental Concepts (page 387) for more information on replication.
897
serverStatus.repl.ismaster The value of the ismaster (page 897) eld is either true or false and reects whether the current node is the master or primary node in the replica set. See Replica Set Fundamental Concepts (page 387) for more information on replication. serverStatus.repl.secondary The value of the secondary (page 898) eld is either true or false and reects whether the current node is a secondary node in the replica set. See Replica Set Fundamental Concepts (page 387) for more information on replication. serverStatus.repl.hosts hosts (page 898) is an array that lists the other nodes in the current replica set. Each member of the replica set appears in the form of hostname:port. See Replica Set Fundamental Concepts (page 387) for more information on replication. opcountersRepl Example output of the opcountersRepl elds. serverStatus.opcountersRepl The opcountersRepl (page 898) data structure, similar to the opcounters (page 899) data structure, provides an overview of database replication operations by type and makes it possible to analyze the load on the replica in more granular manner. These values only appear when the current host has replication enabled. These values will differ from the opcounters (page 899) values because of how MongoDB serializes operations during replication. See Replica Set Fundamental Concepts (page 387) for more information on replication. These numbers will grow over time in response to database use. Analyze these values over time to track database utilization. serverStatus.opcountersRepl.insert insert (page 898) provides a counter of the total number of replicated insert operations since the mongod (page 1021) instance last started. serverStatus.opcountersRepl.query query (page 898) provides a counter of the total number of replicated queries since the mongod (page 1021) instance last started. serverStatus.opcountersRepl.update update (page 898) provides a counter of the total number of replicated update operations since the mongod (page 1021) instance last started. serverStatus.opcountersRepl.delete delete (page 898) provides a counter of the total number of replicated delete operations since the mongod (page 1021) instance last started. serverStatus.opcountersRepl.getmore getmore (page 898) provides a counter of the total number of getmore operations since the mongod (page 1021) instance last started. This counter can be high even if the query count is low. Secondary nodes send getMore operations as part of the replication process. serverStatus.opcountersRepl.command command (page 898) provides a counter of the total number of replicated commands issued to the database since the mongod (page 1021) instance last started.
898
opcounters Example output of the opcounters elds. serverStatus.opcounters The opcounters (page 899) data structure provides an overview of database operations by type and makes it possible to analyze the load on the database in more granular manner. These numbers will grow over time and in response to database use. Analyze these values over time to track database utilization. Note: The data in opcounters (page 899) treats operations that affect multiple documents, such as bulk insert or multi-update operations, as a single operation. See document (page 903) for more granular document-level operation tracking. serverStatus.opcounters.insert insert (page 899) provides a counter of the total number of insert operations since the mongod (page 1021) instance last started. serverStatus.opcounters.query query (page 899) provides a counter of the total number of queries since the mongod (page 1021) instance last started. serverStatus.opcounters.update update (page 899) provides a counter of the total number of update operations since the mongod (page 1021) instance last started. serverStatus.opcounters.delete delete (page 899) provides a counter of the total number of delete operations since the mongod (page 1021) instance last started. serverStatus.opcounters.getmore getmore (page 899) provides a counter of the total number of getmore operations since the mongod (page 1021) instance last started. This counter can be high even if the query count is low. Secondary nodes send getMore operations as part of the replication process. serverStatus.opcounters.command command (page 899) provides a counter of the total number of commands issued to the database since the mongod (page 1021) instance last started. asserts Example output of the asserts elds. serverStatus.asserts The asserts (page 899) document reports the number of asserts on the database. While assert errors are typically uncommon, if there are non-zero values for the asserts (page 899), you should check the log le for the mongod (page 1021) process for more information. In many cases these errors are trivial, but are worth investigating. serverStatus.asserts.regular The regular (page 899) counter tracks the number of regular assertions raised since the server process started. Check the log le for more information about these messages. serverStatus.asserts.warning The warning (page 899) counter tracks the number of warnings raised since the server process started. Check
899
the log le for more information about these warnings. serverStatus.asserts.msg The msg (page 900) counter tracks the number of message assertions raised since the server process started. Check the log le for more information about these messages. serverStatus.asserts.user The user (page 900) counter reports the number of user asserts that have occurred since the last time the server process started. These are errors that user may generate, such as out of disk space or duplicate key. You can prevent these assertions by xing a problem with your application or deployment. Check the MongoDB log for more information. serverStatus.asserts.rollovers The rollovers (page 900) counter displays the number of times that the rollover counters have rolled over since the last time the server process started. The counters will rollover to zero after 230 assertions. Use this value to provide context to the other values in the asserts (page 899) data structure. writeBacksQueued Example output of the writeBacksQueued elds. serverStatus.writeBacksQueued The value of writeBacksQueued (page 900) is true when there are operations from a mongos (page 1032) instance queued for retrying. Typically this option is false. See also: writeBacks Journaling (dur) New in version 1.8. Example output of the journaling elds. serverStatus.dur The dur (page 900) (for durability) document contains data regarding the mongod (page 1021)s journalingrelated operations and performance. mongod (page 1021) must be running with journaling for these data to appear in the output of serverStatus (page 889). Note: The data values are not cumulative but are reset on a regular basis as determined by the journal group commit interval (page 73). This interval is ~100 milliseconds (ms) by default (or 30ms if the journal le is on the same le system as your data les) and is cut by 2/3 when there is a getLastError (page 831) command pending. The interval is congurable using the --journalCommitInterval option. See also: Journaling (page 71) for more information about journaling operations. serverStatus.dur.commits The commits (page 900) provides the number of transactions written to the journal during the last journal group commit interval (page 73). serverStatus.dur.journaledMB The journaledMB (page 900) provides the amount of data in megabytes (MB) written to journal during the last journal group commit interval (page 74).
900
serverStatus.dur.writeToDataFilesMB The writeToDataFilesMB (page 900) provides the amount of data in megabytes (MB) written from journal to the data les during the last journal group commit interval (page 74). serverStatus.dur.compression New in version 2.0. The compression (page 901) represents the compression ratio of the data written to the journal:
( journaled_size_of_data / uncompressed_size_of_data )
serverStatus.dur.commitsInWriteLock The commitsInWriteLock (page 901) provides a count of the commits that occurred while a write lock was held. Commits in a write lock indicate a MongoDB node under a heavy write load and call for further diagnosis. serverStatus.dur.earlyCommits The earlyCommits (page 901) value reects the number of times MongoDB requested a commit before the scheduled journal group commit interval (page 74). Use this value to ensure that your journal group commit interval (page 73) is not too long for your deployment. serverStatus.dur.timeMS The timeMS (page 901) document provides information about the performance of the mongod (page 1021) instance during the various phases of journaling in the last journal group commit interval (page 73). serverStatus.dur.timeMS.dt The dt (page 901) value provides, in milliseconds, the amount of time over which MongoDB collected the timeMS (page 901) data. Use this eld to provide context to the other timeMS (page 901) eld values. serverStatus.dur.timeMS.prepLogBuffer The prepLogBuffer (page 901) value provides, in milliseconds, the amount of time spent preparing to write to the journal. Smaller values indicate better journal performance. serverStatus.dur.timeMS.writeToJournal The writeToJournal (page 901) value provides, in milliseconds, the amount of time spent actually writing to the journal. File system speeds and device interfaces can affect performance. serverStatus.dur.timeMS.writeToDataFiles The writeToDataFiles (page 901) value provides, in milliseconds, the amount of time spent writing to data les after journaling. File system speeds and device interfaces can affect performance. serverStatus.dur.timeMS.remapPrivateView The remapPrivateView (page 901) value provides, in milliseconds, the amount of time spent remapping copy-on-write memory mapped views. Smaller values indicate better journal performance. recordStats Example output of the recordStats elds. serverStatus.recordStats The recordStats (page 901) document provides ne grained reporting on page faults on a per database level. MongoDB uses a read lock on each database to return recordStats (page 901). To minimize this overhead, you can disable this section, as in the following operation:
db.serverStatus( { recordStats: 0 } )
serverStatus.recordStats.accessesNotInMemory accessesNotInMemory (page 901) reects the number of times mongod (page 1021) needed to access a memory page that was not resident in memory for all databases managed by this mongod (page 1021) instance.
901
serverStatus.recordStats.pageFaultExceptionsThrown pageFaultExceptionsThrown (page 901) reects the number of page fault exceptions thrown by mongod (page 1021) when accessing data for all databases managed by this mongod (page 1021) instance. serverStatus.recordStats.local.accessesNotInMemory accessesNotInMemory (page 902) reects the number of times mongod (page 1021) needed to access a memory page that was not resident in memory for the local database. serverStatus.recordStats.local.pageFaultExceptionsThrown pageFaultExceptionsThrown (page 902) reects the number of page fault exceptions thrown by mongod (page 1021) when accessing data for the local database. serverStatus.recordStats.admin.accessesNotInMemory accessesNotInMemory (page 902) reects the number of times mongod (page 1021) needed to access a memory page that was not resident in memory for the admin database. serverStatus.recordStats.admin.pageFaultExceptionsThrown pageFaultExceptionsThrown (page 902) reects the number of page fault exceptions thrown by mongod (page 1021) when accessing data for the admin database. serverStatus.recordStats.<database>.accessesNotInMemory accessesNotInMemory (page 902) reects the number of times mongod (page 1021) needed to access a memory page that was not resident in memory for the <database> database. serverStatus.recordStats.<database>.pageFaultExceptionsThrown pageFaultExceptionsThrown (page 902) reects the number of page fault exceptions thrown by mongod (page 1021) when accessing data for the <database> database. workingSet New in version 2.4. Example output of the workingSet elds. Note: The workingSet (page 902) data is only included in the output of serverStatus (page 889) if explicitly enabled. To return the workingSet (page 902) use one of the following commands:
db.serverStatus( { workingSet: 1 } ) db.runCommand( { serverStatus: 1, workingSet: 1 } )
serverStatus.workingSet workingSet (page 902) is a document that contains values useful for estimating the size of the working set, which is the amount of data that MongoDB uses actively. workingSet (page 902) uses an internal data structure that tracks pages accessed by mongod (page 1021). serverStatus.workingSet.note note (page 902) is a eld that holds a string warning that the workingSet (page 902) document is an estimate. serverStatus.workingSet.pagesInMemory pagesInMemory (page 902) contains a count of the total number of pages accessed by mongod (page 1021) over the period displayed in overSeconds (page 903). The default page size is 4 kilobytes: to convert this value to the amount of data in memory multiply this value by 4 kilobytes. If your total working set is less than the size of physical memory, over time the value of pagesInMemory (page 902) will reect your data size.
902
Use pagesInMemory (page 902) in conjunction with overSeconds (page 903) to help estimate the actual size of the working set. serverStatus.workingSet.computationTimeMicros computationTimeMicros (page 903) reports the amount of time the mongod (page 1021) instance used to compute the other elds in the workingSet (page 902) section. Reporting on workingSet (page 902) may impact the performance of other operations on the mongod (page 1021) instance because MongoDB must collect some data within the context of a lock. Ensure that automated monitoring tools consider this metric when determining the frequency of collection for workingSet (page 902). serverStatus.workingSet.overSeconds overSeconds (page 903) returns the amount of time elapsed between the newest and oldest pages tracked in the pagesInMemory (page 902) data point. If overSeconds (page 903) is decreasing, or if pagesInMemory (page 902) equals physical RAM and overSeconds (page 903) is very small, the working set may be much larger than physical RAM. When overSeconds (page 903) is large, MongoDBs data set is equal to or smaller than physical RAM. metrics Example output of the metrics elds. New in version 2.4. serverStatus.metrics The metrics (page 903) document holds a number of statistics that reect the current use and state of a running mongod (page 1021) instance. serverStatus.metrics.document The document (page 903) holds a document of that reect document access and modication patterns and data use. Compare these values to the data in the opcounters (page 899) document, which track total number of operations. serverStatus.metrics.document.deleted deleted (page 903) reports the total number of documents deleted. serverStatus.metrics.document.inserted inserted (page 903) reports the total number of documents inserted. serverStatus.metrics.document.returned returned (page 903) reports the total number of documents returned by queries. serverStatus.metrics.document.updated updated (page 903) reports the total number of documents updated. serverStatus.metrics.getLastError getLastError (page 903) is a document that reports on getLastError (page 831) use. serverStatus.metrics.getLastError.wtime wtime (page 903) is a sub-document that reports getLastError (page 831) operation counts with a w argument greater than 1. serverStatus.metrics.getLastError.wtime.num num (page 903) reports the total number of getLastError (page 831) operations without a specied write concern (i.e. w) that wait for one or more members of a replica set to acknowledge the write operation (i.e. greater than 1.)
903
serverStatus.metrics.getLastError.wtime.totalMillis totalMillis (page 903) reports the total amount of time in milliseconds that the mongod (page 1021) has spent performing getLastError (page 831) operations with write concern (i.e. w) that wait for one or more members of a replica set to acknowledge the write operation (i.e. greater than 1.) serverStatus.metrics.getLastError.wtimeouts wtimeouts (page 904) reports the number of times that write concern operations have timed out as a result of the wtimeout threshold to getLastError (page 831). serverStatus.metrics.operation operation (page 904) is a sub-document that holds counters for several types of update and query operations that MongoDB handles using special operation types. serverStatus.metrics.operation.fastmod fastmod (page 904) reports the number of update (page 219) operations that neither cause documents to grow nor require updates to the index. For example, this counter would record an update operation that use the $inc (page 784) operator to increment the value of a eld that is not indexed. serverStatus.metrics.operation.idhack idhack (page 904) reports the number of queries that contain the _id eld. For these queries, MongoDB will use default index on the _id eld and skip all query plan analysis. serverStatus.metrics.operation.scanAndOrder scanAndOrder (page 904) reports the total number of queries that return sorted numbers that cannot perform the sort operation using an index. serverStatus.metrics.queryExecutor queryExecutor (page 904) is a document that reports data from the query execution system. serverStatus.metrics.queryExecutor.scanned scanned (page 904) reports the total number of index items scanned during queries and query-plan evaluation. This counter is the same as nscanned (page 956) in the output of explain() (page 953). serverStatus.metrics.record record (page 904) is a document that reports data related to record allocation in the on-disk memory les. serverStatus.metrics.record.moves moves (page 904) reports the total number of times documents move within the on-disk representation of the MongoDB data set. Documents move as a result of operations that increase the size of the document beyond their allocated record size. serverStatus.metrics.repl repl (page 904) holds a sub-document that reports metrics related to the replication process. repl (page 904) document appears on all mongod (page 1021) instances, even those that arent members of replica sets. serverStatus.metrics.repl.apply apply (page 904) holds a sub-document that reports on the application of operations from the replication oplog. serverStatus.metrics.repl.apply.batches batches (page 904) reports on the oplog application process on secondaries members of replica sets. See Multithreaded Replication (page 411) for more information on the oplog application processes serverStatus.metrics.repl.apply.batches.num num (page 904) reports the total number of batches applied across all databases. serverStatus.metrics.repl.apply.batches.totalMillis totalMillis (page 904) reports the total amount of time the mongod (page 1021) has spent applying operations from the oplog. serverStatus.metrics.repl.apply.ops ops (page 904) reports the total number of oplog operations applied.
904
serverStatus.metrics.repl.buffer MongoDB buffers oplog operations from the replication sync source buffer before applying oplog entries in a batch. buffer (page 904) provides a way to track the oplog buffer. See Multithreaded Replication (page 411) for more information on the oplog application process. serverStatus.metrics.repl.buffer.count count (page 905) reports the current number of operations in the oplog buffer. serverStatus.metrics.repl.buffer.maxSizeBytes maxSizeBytes (page 905) reports the maximum size of the buffer. This value is a constant setting in the mongod (page 1021), and is not congurable. serverStatus.metrics.repl.buffer.sizeBytes sizeBytes (page 905) reports the current size of the contents of the oplog buffer. serverStatus.metrics.repl.network network (page 905) reports network use by the replication process. serverStatus.metrics.repl.network.bytes bytes (page 905) reports the total amount of data read from the replication sync source. serverStatus.metrics.repl.network.getmores getmores (page 905) reports on the getmore operations,which are requests for additional results from the oplog cursor as part of the oplog replication process. serverStatus.metrics.repl.network.getmores.num num (page 905) reports the total number of getmore operations, which are operations that request an additional set of operations from the replication sync source. serverStatus.metrics.repl.network.getmores.totalMillis totalMillis (page 905) reports the total amount of time required to collect data from getmore operations. Note: This number can be quite large, as MongoDB will wait for more data even if the getmore operation does not initial return data. serverStatus.metrics.repl.network.ops ops (page 905) reports the total number of operations read from the replication source. serverStatus.metrics.repl.network.readersCreated readersCreated (page 905) reports the total number of oplog query processes created. MongoDB will create a new oplog query any time an error occurs in the connection, including a timeout, or a network operation. Furthermore, readersCreated (page 905) will increment every time MongoDB selects a new source fore replication. serverStatus.metrics.repl.oplog oplog (page 905) is a document that reports on the size and use of the oplog by this mongod (page 1021) instance. serverStatus.metrics.repl.oplog.insert insert (page 905) is a document that reports insert operations into the oplog. serverStatus.metrics.repl.oplog.insert.num num (page 905) reports the total number of items inserted into the oplog. serverStatus.metrics.repl.oplog.insert.totalMillis totalMillis (page 905) reports the total amount of time spent for the mongod (page 1021) to insert data into the oplog. serverStatus.metrics.repl.oplog.insertBytes insertBytes (page 905) the total size of documents inserted into the oplog. 63.2. Database Commands 905
serverStatus.metrics.repl.preload preload (page 905) reports on the pre-fetch stage, where MongoDB loads documents and indexes into RAM to improve replication throughput. See Multithreaded Replication (page 411) for more information about the pre-fetch stage of the replication process. serverStatus.metrics.repl.preload.docs docs (page 906) is a sub-document that reports on the documents loaded into memory during the pre-fetch stage. serverStatus.metrics.repl.preload.docs.num num (page 906) reports the total number of documents loaded during the pre-fetch stage of replication. serverStatus.metrics.repl.preload.docs.totalMillis totalMillis (page 906) reports the total amount of time spent loading documents as part of the pre-fetch stage of replication. serverStatus.metrics.repl.preload.indexes indexes (page 906) is a sub-document that reports on the index items loaded into memory during the pre-fetch stage of replication. See Multithreaded Replication (page 411) for more information about the pre-fetch stage of replication. serverStatus.metrics.repl.preload.indexes.num num (page 906) reports the total number of index entries loaded by members before updating documents as part of the pre-fetch stage of replication. serverStatus.metrics.repl.preload.indexes.totalMillis totalMillis (page 906) reports the total amount of time spent loading index entries as part of the pre-fetch stage of replication. serverStatus.metrics.ttl ttl (page 906) is a sub-document that reports on the operation of the resource use of the ttl index (page 577) process. serverStatus.metrics.ttl.deletedDocuments deletedDocuments (page 906) reports the total number of documents deleted from collections with a ttl index (page 577). serverStatus.metrics.ttl.passes passes (page 906) reports the number of times the background process removes documents from collections with a ttl index (page 577). top top The top (page 906) command is an administrative command which returns raw usage of each database, and provides amount of time, in microseconds, used and a count of operations for the following event types: total readLock writeLock queries getmore insert update
906
remove commands You must issue the top (page 906) command against the admin database in the form:
{ top: 1 }
validate Denition validate The validate (page 907) command checks the structures within a namespace for correctness by scanning the collections data and indexes. The command returns information regarding the on-disk representation of the collection. The validate command can be slow, particularly on larger data sets. The following example validates the contents of the collection named users.
{ validate: "users" }
You may also specify one of the following options: full: true provides a more thorough scan of the data. false skips the scan of the base collection without skipping the scan of the index.
scandata:
The mongo (page 1036) shell also provides a wrapper:

db.collection.validate();
Use one of the following forms to perform the full collection validation:
db.collection.validate(true) db.runCommand( { validate: "collection", full: true } )
Warning: This command is resource intensive and may have an impact on the performance of your MongoDB instance.
Output validate.ns The full namespace name of the collection. Namespaces include the database name and the collection name in the form database.collection. validate.firstExtent The disk location of the rst extent in the collection. The value of this eld also includes the namespace. validate.lastExtent The disk location of the last extent in the collection. The value of this eld also includes the namespace. validate.extentCount The number of extents in the collection. validate.extents validate (page 907) returns one instance of this document for every extent in the collection. This subdocument is only returned when you specify the full option to the command. validate.extents.loc The disk location for the beginning of this extent. 63.2. Database Commands 907
validate.extents.xnext The disk location for the extent following this one. null if this is the end of the linked list of extents. validate.extents.xprev The disk location for the extent preceding this one. null if this is the head of the linked list of extents. validate.extents.nsdiag The namespace this extent belongs to (should be the same as the namespace shown at the beginning of the validate listing). validate.extents.size The number of bytes in this extent. validate.extents.firstRecord The disk location of the rst record in this extent. validate.extents.lastRecord The disk location of the last record in this extent. validate.datasize The number of bytes in all data records. This value does not include deleted records, nor does it include extent headers, nor record headers, nor space in a le unallocated to any extent. datasize (page 908) includes record padding. validate.nrecords The number of documents in the collection. validate.lastExtentSize The size of the last new extent created in this collection. This value determines the size of the next extent created. validate.padding A oating point value between 1 and 2. When MongoDB creates a new record it uses the padding factor to determine how much additional space to add to the record. The padding factor is automatically adjusted by mongo when it notices that update operations are triggering record moves. validate.firstExtentDetails The size of the rst extent created in this collection. This data is similar to the data provided by the extents (page 907) sub-document; however, the data reects only the rst extent in the collection and is always returned. validate.firstExtentDetails.loc The disk location for the beginning of this extent. validate.firstExtentDetails.xnext The disk location for the extent following this one. null if this is the end of the linked list of extents, which should only be the case if there is only one extent. validate.firstExtentDetails.xprev The disk location for the extent preceding this one. This should always be null. validate.firstExtentDetails.nsdiag The namespace this extent belongs to (should be the same as the namespace shown at the beginning of the validate listing). validate.firstExtentDetails.size The number of bytes in this extent. validate.firstExtentDetails.firstRecord The disk location of the rst record in this extent.
908
validate.firstExtentDetails.lastRecord The disk location of the last record in this extent. validate.objectsFound The number of records actually encountered in a scan of the collection. This eld should have the same value as the nrecords (page 908) eld. validate.invalidObjects The number of records containing BSON documents that do not pass a validation check. Note: This eld is only included in the validation output when you specify the full option. validate.bytesWithHeaders This is similar to datasize, except that bytesWithHeaders (page 909) includes the record headers. In version 2.0, record headers are 16 bytes per document. Note: This eld is only included in the validation output when you specify the full option. validate.bytesWithoutHeaders bytesWithoutHeaders (page 909) returns data collected from a scan of all records. The value should be the same as datasize (page 908). Note: This eld is only included in the validation output when you specify the full option. validate.deletedCount The number of deleted or free records in the collection. validate.deletedSize The size of all deleted or free records in the collection. validate.nIndexes The number of indexes on the data in the collection. validate.keysPerIndex A document containing a eld for each index, named after the indexs name, that contains the number of keys, or documents referenced, included in the index. validate.valid Boolean. true, unless validate (page 907) determines that an aspect of the collection is not valid. When false, see the errors (page 909) eld for more information. validate.errors Typically empty; however, if the collection is not valid (i.e valid (page 909) is false,) this eld will contain a message describing the validation error. validate.ok Set to 1 when the command succeeds. If the command fails the ok (page 909) eld has a value of 0. whatsmyuri whatsmyuri whatsmyuri (page 909) is an internal command.
909
63.2.3 Internal Commands

Internal Commands Name _migrateClone (page 910) _recvChunkAbort (page 910) _recvChunkCommit (page 910) _recvChunkStart (page 910) _recvChunkStatus (page 911) _replSetFresh _transferMods (page 911) handshake (page 911) mapreduce.shardedfinish (page 911) replSetElect (page 911) replSetGetRBID (page 911) replSetHeartbeat (page 911) writeBacksQueued (page 912) writebacklisten (page 912)
migrateClone
Description Internal command that supports chunk migration. Do not call directly. Internal command that supports chunk migrations in sharded clusters. Do not call directly. Internal command that supports chunk migrations in sharded clusters. Do not call directly. Internal command that facilitates chunk migrations in sharded clusters.. Do not call directly. Internal command that returns data to support chunk migrations in sharded clusters. Do not call directly. Internal command that supports replica set election operations. Internal command that supports chunk migrations. Do not call directly. Internal command. Internal command that supports map-reduce in sharded cluster environments. Internal command that supports replica set functionality. Internal command that supports replica set operations. Internal command that supports replica set operations. Internal command that supports chunk migrations in sharded clusters. Internal command that supports chunk migrations in sharded clusters.
_migrateClone _migrateClone (page 910) is an internal command. Do not call directly.

recvChunkAbort
_recvChunkAbort _recvChunkAbort (page 910) is an internal command. Do not call directly.

recvChunkCommit
_recvChunkCommit _recvChunkCommit (page 910) is an internal command. Do not call directly.

recvChunkStart
_recvChunkStart _recvChunkStart (page 910) is an internal command. Do not call directly. Warning: This command obtains a write lock on the affected database and will block other operations until it has completed.
910
recvChunkStatus
_recvChunkStatus _recvChunkStatus (page 911) is an internal command. Do not call directly.

replSetFresh
replSetFresh replSetFresh (page 911) is an internal command that supports replica set functionality.
transferMods
_transferMods _transferMods (page 911) is an internal command. Do not call directly.

handshake
handshake handshake (page 911) is an internal command.

mapreduce.shardednish
mapreduce.shardedfinish Provides internal functionality to support map-reduce in sharded environments. See also: mapReduce (page 814)
replSetElect
replSetElect replSetElect (page 911) is an internal command that support replica set functionality.
replSetGetRBID
replSetGetRBID replSetGetRBID (page 911) is an internal command that supports replica set functionality.
replSetHeartbeat
replSetHeartbeat replSetHeartbeat (page 911) is an internal command that supports replica set functionality.
911
writeBacksQueued
writeBacksQueued writeBacksQueued (page 912) is an internal command that returns a document reporting there are operations in the write back queue for the given mongos (page 1032) and information about the queues. writeBacksQueued.hasOpsQueued Boolean. hasOpsQueued (page 912) is true if there are write Back operations queued. writeBacksQueued.totalOpsQueued Integer. totalOpsQueued (page 912) reects the number of operations queued. writeBacksQueued.queues Document. queues (page 912) holds a sub-document where the elds are all write back queues. These eld hold a document with two elds that reports on the state of the queue. The elds in these documents are: writeBacksQueued.queues.n n (page 912) reects the size, by number of items, in the queues. writeBacksQueued.queues.minutesSinceLastCall The number of minutes since the last time the mongos (page 1032) touched this queue. The command document has the following prototype form:
{writeBacksQueued: 1}
To call writeBacksQueued (page 912) from the mongo (page 1036) shell, use the following db.runCommand() (page 984) form:
db.runCommand({writeBacksQueued: 1})
Consider the following example output:

{ "hasOpsQueued" : true, "totalOpsQueued" : 7, "queues" : { "50b4f09f6671b11ff1944089" "50b4f09fc332bf1c5aeaaf59" "50b4f09f6671b1d51df98cb6" "50b4f0c67ccf1e5c6effb72e" "50b4faf12319f193cfdec0d1" "50b4f013d2c1f8d62453017e" "50b4f0f12319f193cfdec0d1" }, "ok" : 1 }
: : : : : : :
{ { { { { { {
"n" "n" "n" "n" "n" "n" "n"
: : : : : : :
0, 0, 0, 0, 0, 0, 0,
"minutesSinceLastCall" "minutesSinceLastCall" "minutesSinceLastCall" "minutesSinceLastCall" "minutesSinceLastCall" "minutesSinceLastCall" "minutesSinceLastCall"
: : : : : : :
1 0 0 0 4 0 1
}, }, }, }, }, }, }
writebacklisten
writebacklisten writebacklisten (page 912) is an internal command.
912
63.2.4 Testing Commands

Testing Commands Name _hashBSONElement (page 913) _journalLatencyTest Description Internal command. Computes the MD5 hash of a BSON element. Tests the time required to write and perform a le system sync for a le in the journal directory. Internal command. Truncates capped collections. Internal command for testing. Congures failure points.
captrunc (page 914) configureFailPoint (page 915) emptycapped (page 915) Internal command. Removes all documents from a capped collection. forceerror (page 916) Internal command for testing. Forces a user assertion exception. godinsert (page 916) Internal command for testing. replSetTest (page 916) Internal command for testing replica set functionality. skewClockCommand Internal command. Do not call this command directly. sleep (page 916) Internal command for testing. Forces MongoDB to block all operations. testDistLockWithSkew Internal command. Do not call this directly. testDistLockWithSyncCluster Internal command. Do not call this directly.
_hashBSONElement
_hashBSONElement New in version 2.4. _hashBSONElement (page 913) is an internal command that computes the MD5 hash of a BSON element. _hashBSONElement (page 913) returns 8 bytes from the 16 byte MD5 hash. Parameters key (BSONElement) Species the BSON element to hash. seed (integer) Species a seed to use to compute the hash. Returns A document consisting of key, seed, out and ok elds. Fields key (BSONType) The original BSON element. seed (integer) The seed used for the hash, defaults to 0. out (NumberLong) The decimal result of the hash. ok (integer) Holds the 1 if the function returns successfully, and 0 if the operation encountered an error. Note: _hashBSONElement (page 913) is an internal command that is not enabled by default. _hashBSONElement (page 913) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. _hashBSONElement (page 913) cannot be enabled during run-time. Example
mongod --setParameter=enableTestCommands=1
913
Example Hash an ISODate string:

db.runCommand({_hashBSONElement: ISODate("2013-02-12T22:12:57.211Z")})
Which returns the following document:

{ "key" : ISODate("2013-02-12T22:12:57.211Z"), "seed" : 0, "out" : NumberLong("-4185544074338741873"), "ok" : 1 }
Hash the same ISODate string but specify a seed value:

db.runCommand({_hashBSONElement: ISODate("2013-02-12T22:12:57.211Z"), seed:2013})
Which returns the following document:

{ "key" : ISODate("2013-02-12T22:12:57.211Z"), "seed" : 2013, "out" : NumberLong("7845924651247493302"), "ok" : 1 }
journalLatencyTest
journalLatencyTest journalLatencyTest (page 914) is an administrative command that tests the length of time required to write and perform a le system sync (e.g. fsync) for a le in the journal directory. You must issue the journalLatencyTest (page 914) command against the admin database in the form:
{ journalLatencyTest: 1 }
The value (i.e. 1 above), does not affect the operation of the command. Note: journalLatencyTest (page 914) is an internal command that is not enabled by default. journalLatencyTest (page 914) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. journalLatencyTest (page 914) cannot be enabled during run-time.
captrunc
captrunc The captrunc (page 914) command is an internal command that truncates capped collections. Parameters collection (string) The name of the collection to truncate. n (integer) An integer that species the number of documents to remove from the collection 914 Chapter 63. MongoDB Interface
inc (boolean) Species whether to truncate the nth document. Example Truncate 10 older documents from the collection records:
db.runCommand({captrunc: "records" , n:10})
Truncate 100 documents and the 101st document:

db.runCommand({captrunc: "records", n:100, inc:true})
Note: captrunc (page 914) is an internal command that is not enabled by default. captrunc (page 914) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. captrunc (page 914) cannot be enabled during run-time.
congureFailPoint
configureFailPoint configureFailPoint (page 915) is an internal testing command that congures failure points. Parameters string Species the name of the failure point. mode (string) Controls behavior of failure point. Possible values are: alwaysOn and off. mode Specify a document {times: will activate the failure point. n} to control the number of times that MongoDB
Note: configureFailPoint (page 915) is an internal command that is not enabled by default. configureFailPoint (page 915) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. configureFailPoint (page 915) cannot be enabled during run-time.
emptycapped
emptycapped The emptycapped command removes all documents from a capped collection. Use the following syntax:
{ emptycapped: "events" }
This command removes all records from the capped collection named events. Warning: This command obtains a write lock on the affected database and will block other operations until it has completed.
Note: emptycapped (page 915) is an internal command that is not enabled by default. emptycapped (page 915) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. emptycapped (page 915) cannot be enabled during run-time.
915
forceerror
forceerror The forceerror (page 916) command is for testing purposes only. Use forceerror (page 916) to force a user assertion exception. This command always returns an ok value of 0.
godinsert
godinsert godinsert (page 916) is an internal command for testing purposes only. Note: This command obtains a write lock on the affected database and will block other operations until it has completed. Note: godinsert (page 916) is an internal command that is not enabled by default. godinsert (page 916) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. godinsert (page 916) cannot be enabled during run-time.
replSetTest
replSetTest replSetTest (page 916) is internal diagnostic command used for regression tests that supports replica set functionality. Note: replSetTest (page 916) is an internal command that is not enabled by default. replSetTest (page 916) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. replSetTest (page 916) cannot be enabled during run-time.
skewClockCommand
_skewClockCommand _skewClockCommand (page 916) is an internal command. Do not call directly. Note: _skewClockCommand (page 916) is an internal command that is not enabled by default. _skewClockCommand (page 916) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. _skewClockCommand (page 916) cannot be enabled during run-time.
sleep
sleep sleep (page 916) is an internal command for testing purposes. The sleep (page 916) command forces the database to block all operations. It takes the following options: Parameters w (boolean) If true, obtain a global write lock. Otherwise obtains a read lock. secs (integer) Species the number of seconds to sleep.
916
{ sleep: { w: true, secs: <seconds> } }
The above command places the mongod (page 1021) instance in a write-lock state for a specied (i.e. <seconds>) number of seconds. Without arguments, sleep (page 916), causes a read lock for 100 seconds. Warning: sleep (page 916) claims the lock specied in the w argument and blocks all operations on the mongod (page 1021) instance for the specied amount of time.
Note: sleep (page 916) is an internal command that is not enabled by default. sleep (page 916) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. sleep (page 916) cannot be enabled during run-time.
testDistLockWithSkew
_testDistLockWithSkew _testDistLockWithSkew (page 917) is an internal command. Do not call directly. Note: _testDistLockWithSkew (page 917) is an internal command that is not enabled by default. _testDistLockWithSkew (page 917) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. _testDistLockWithSkew (page 917) cannot be enabled during run-time.
testDistLockWithSyncCluster
_testDistLockWithSyncCluster _testDistLockWithSyncCluster (page 917) is an internal command. Do not call directly. Note: _testDistLockWithSyncCluster (page 917) is an internal command that is not enabled by default. _testDistLockWithSyncCluster (page 917) must be enabled by using --setParameter enableTestCommands=1 (page 1027) on the mongod (page 1021) command line. _testDistLockWithSyncCluster (page 917) cannot be enabled during run-time.
63.3 mongo Shell Methods

Collection (page 918) Cursor (page 951) Database (page 966) Replication (page 987) Sharding (page 993) Connection (page 1007) Subprocess (page 1010) Native (page 1012)
JavaScript in MongoDB
63.3. mongo Shell Methods
917
Although these methods use JavaScript, most interactions with MongoDB do not use JavaScript but use an idiomatic driver (page 555) in the language of the interacting application.
63.3.1 Collection
Collection Methods Name db.collection.aggregate() (page 918) db.collection.count() (page 919) db.collection.createIndex() (page 920) db.collection.dataSize() (page 920) db.collection.distinct() (page 920) db.collection.drop() (page 921) db.collection.dropIndex() db.collection.dropIndexes() (page 921) db.collection.ensureIndex (page 921) db.collection.find() (page 924) db.collection.findAndModify() (page 926) db.collection.findOne() (page 929) db.collection.getIndexStats() (page 929) db.collection.getIndexes() (page 930) db.collection.getShardDistribution() (page 931) db.collection.getShardVersion() (page 932) db.collection.group() (page 932) db.collection.insert() (page 936) db.collection.isCapped() (page 937) db.collection.mapReduce() (page 937) db.collection.reIndex() (page 944) db.collection.remove() (page 944) db.collection.renameCollection() (page 945) db.collection.save() (page 946) db.collection.stats() (page 947) db.collection.storageSize() (page 947) db.collection.storageSize() (page 947) db.collection.totalSize() (page 948) db.collection.update() (page 948) db.collection.validate() (page 950)
db.collection.aggregate()
Description Provides access to the aggregation framework (page 255) pipeline Wraps count (page 808) to return a count of the number of docu Builds an index on a collection. Use db.collection.ensure Returns the size of the collection. Wraps the size (page 873) e Returns an array of documents that have distinct values for the spe Removes the specied collection from the database. Removes a specied index on a collection. Removes all indexes on a collection. Creates an index if it does not currently exist. If the index exists e Performs a query on a collection and returns a cursor object. Atomically modies and returns a single document. Performs a query and returns a single document. Renders a human-readable view of the data collected by indexS Returns an array of documents that describe the existing indexes o For collections in sharded clusters, db.collection.getShar Internal diagnostic method for shard cluster. Provides simple data aggregation function. Groups documents in Creates a new document in a collection. Reports if a collection is a capped collection. Performs map-reduce style data aggregation. Rebuilds all existing indexes on a collection. Deletes documents from a collection. Changes the name of a collection. Provides a wrapper around an insert() (page 936) and updat Reports on the state of a collection. Provides a wrapper around th Reports the total size used by the collection. Provides a wrapper a Reports the total size used by the indexes on a collection. Provide Reports the total size of a collection, including the size of all docu Modies a document in a collection. Performs diagnostic operations on a collection.
db.collection.aggregate(pipeline) New in version 2.2. Always call the db.collection.aggregate() (page 918) method on a collection object. Arguments pipeline Species a sequence of data aggregation processes. See the aggregation reference (page 271) for documentation of these operators. Returns 918 Chapter 63. MongoDB Interface
A document with two elds: result which holds an array of documents returned by the pipeline ok which holds the value 1, indicating success. Throws exception Changed in version 2.4: If an error occurs, the aggregate() (page 918) helper throws an exception. In previous versions, the helper returned a document with the error message and code, and ok status eld not equal to 1, same as the aggregate (page 808) command. Consider the following example from the aggregation documentation (page 255).
db.article.aggregate( { $project : { author : 1, tags : 1, } }, { $unwind : "$tags" }, { $group : { _id : { tags : 1 }, authors : { $addToSet : "$author" } } } );
See also: aggregate (page 808), Aggregation Framework (page 255), and Aggregation Framework Reference (page 271).
db.collection.count()
db.collection.count() The db.collection.count() (page 919) method is a shell wrapper that returns the count of documents that would match a find() (page 924) query; i.e., db.collection.count() (page 919) method is equivalent to:
db.collection.find(<query>).count();
This operation does not actually perform the find() (page 924); instead, the operation counts the results that would be returned by the find() (page 924). The db.collection.count() (page 919) method can accept the following argument: Parameters query (document) Species the selection query criteria. Consider the following examples of the db.collection.count() (page 919) method Count the number of all documents in the orders collection:
db.orders.count()
The query is equivalent to the following:

db.orders.find().count()
919
db.orders.count( { ord_dt: { $gt: new Date(01/01/2012) } } )
The query is equivalent to the following:

db.orders.find( { ord_dt: { $gt: new Date(01/01/2012) } } ).count()
See also: cursor.count() (page 952)

db.collection.createIndex()
db.collection.createIndex(keys, options) Deprecated since version 1.8. Parameters keys (document) A document that contains pairs with the name of the eld or elds to index and order of the index. A 1 species ascending and a -1 species descending. options (document) A document that controls the creation of the index. This argument is optional. The ensureIndex() (page 921) method is the preferred way to create indexes on collections. See also: Indexes (page 327), db.collection.createIndex() (page 920), db.collection.dropIndex(), db.collection.dropIndexes() (page 921), db.collection.getIndexes() (page 930), db.collection.reIndex() (page 944), and db.collection.totalIndexSize()
db.collection.dataSize()
db.collection.dataSize() Returns The size of the collection. This method provides a wrapper around the size (page 873) output of the collStats (page 873) (i.e. db.collection.stats() (page 947)) command.
db.collection.distinct()
db.collection.distinct() The db.collection.distinct() (page 920) method nds the distinct values for a specied eld across a single collection and returns the results in an array. The method accepts the following argument: Parameters eld (string) Species the eld for which to return the distinct values. query (document) Species the selection query to determine the subset of documents from which to retrieve the distinct values. Note: The db.collection.distinct() (page 920) method provides a wrapper around the distinct (page 809) command. Results must not be larger than the maximum BSON size (page 1105).
920
When possible to use covered indexes, the db.collection.distinct() (page 920) method will use an index to nd the documents in the query as well as to return the data. Consider the following examples of the db.collection.distinct() (page 920) method: Return an array of the distinct values of the eld ord_dt from all documents in the orders collection:
db.orders.distinct( ord_dt )
Return an array of the distinct values of the eld sku in the subdocument item from all documents in the orders collection:
db.orders.distinct( item.sku )
Return an array of the distinct values of the eld ord_dt from the documents in the orders collection where the price is greater than 10:
db.orders.distinct( ord_dt, { price: { $gt: 10 } } )
db.collection.drop()
db.collection.drop() Call the db.collection.drop() (page 921) method on a collection to drop it from the database. db.collection.drop() (page 921) takes no arguments and will produce an error if called with any arguments.
db.collection.dropIndexes()
db.collection.dropIndexes() Drops all indexes other than the required index on the _id eld. Only call dropIndexes() (page 921) as a method on a collection object.
db.collection.dropIndexes()
db.collection.dropIndexes() Drops all indexes other than the required index on the _id eld. Only call dropIndexes() (page 921) as a method on a collection object.
db.collection.ensureIndex()
db.collection.ensureIndex(keys, options) Creates an index on the eld specied, if that index does not already exist. Parameters keys (document) For ascending/descending indexes, a document (page 192) that contains pairs with the name of the eld or elds to index and order of the index. A 1 species ascending and a -1 species descending. MongoDB supports several different index types including text (page 338), geospatial (page 357), and hashed (page 335) indexes.
921
options (document) A document that controls the creation of the index. This argument is optional. Warning: Index names, including their full namespace (i.e. database.collection) cannot be longer than 128 characters. See the db.collection.getIndexes() (page 930) eld name (page 930) for the names of existing indexes.
Example Create an Ascending Index on a Single Field The following example creates an ascending index on the eld orderDate.
db.collection.ensureIndex( { orderDate: 1 } )
If the keys document species more than one eld, then db.collection.ensureIndex() (page 921) creates a compound index. Example Create an Index on a Multiple Fields The following example creates a compound index on the orderDate eld (in ascending order) and the zipcode eld (in descending order.)
db.collection.ensureIndex( { orderDate: 1, zipcode: -1 } )
A compound index cannot include a hashed index (page 335) component. Note: The order of an index is important for supporting cursor.sort() (page 965) operations using the index. See also: The Indexes (page 327) section of this manual for full documentation of indexes and indexing in MongoDB. The Create a text Index (page 372) section for more information and examples on creating text indexes. ensureIndex() (page 921) provides the following options:
922
Option
Value
Default
background true false or false unique true false or false name string none dropDups true false or false sparse true false or false expireAfterSeconds integer none v index The default index version depends on the version of mongod ver(page 1021) running when creating the index. Before version 2.0, the sion this value was 0; versions 2.0 and later use version 1. weights docu1 ment default_language string english language_override string language Options
Index Type All
Ascending/Descending All Scalar
Ascending/Descending TTL All
Text Text Text
background (boolean) Specify true to build the index in the background so that building an index will not block other database activities. unique (boolean) Specify true to create a unique index so that the collection will not accept insertion of documents where the index key or keys matches an existing value in the index. name (string) Specify the name of the index. If unspecied, MongoDB will generate an index name by concatenating the names of the indexed elds and the sort order. dropDups (boolean) Specify true when creating a unique index, on a eld that may have duplicate to index only the rst occurrence of a key, and remove all documents from the collection that contain subsequent occurrences of that key. sparse (boolean) If true, the index only references documents with the specied eld. These indexes use less space, but behave differently in some situations (particularly sorts.) expireAfterSeconds (integer) Specify a value, in seconds, as a TTL to control how long MongoDB will retain documents in this collection. See Expire Data from Collections by Setting TTL (page 577) for more information on this functionality. v Only specify a different index version in unusual situations. The latest index version (version 1) provides a smaller and faster index format. weights (document) For text index only. The document contains eld and weight pairs. The weight is a number ranging from 1 to 99,999. The weight of the index eld denote the signicance of the eld relative to the other indexed elds in terms of the score. You can specify weights for some or all the indexed elds. See Control Results of Text Search with Weights (page 375) to adjust the scores. default_language (string) For text index only. Specify the language that determines the list of stop words and the rules for the stemmer and tokenizer. The default language for the 63.3. mongo Shell Methods 923
indexed data is english. See Text Search Languages (page 836) for the available languages and Specify a Language for Text Index (page 373) for more information and example. language_override (string) For text index only. Specify the name of the eld in the document that contains, for that document, the language to override the default language. See Create a text Index on a Multi-language Collection (page 374). Please be aware of the following behaviors of ensureIndex() (page 921): To add or change index options you must drop the index using the dropIndex() method and issue another ensureIndex() (page 921) operation with the new options. If you create an index with one set of options, and then issue the ensureIndex() (page 921) method with the same index elds and different options without rst dropping the index, ensureIndex() (page 921) will not rebuild the existing index with the new options. If you call multiple ensureIndex() (page 921) methods with the same index specication at the same time, only the rst operation will succeed, all other operations will have no effect. Non-background indexing operations will block all other operations on a database. See also: In addition to the ascending/descending indexes, MongoDB provides the following index types to provide additional functionalities: TTL Indexes (page 337) to support expiration of data, Geospatial Indexes (page 338) and Geohaystack Indexes (page 338) to support geospatial queries, and text Indexes (page 338) to support text searches.
db.collection.nd()
db.collection.find(query, projection) The find() (page 924) method selects documents in a collection and returns a cursor to the selected documents. The find() (page 924) method takes the following parameters. Parameters query (document) Optional. Species the selection criteria using query operators (page 763). Omit the query parameter or pass an empty document (e.g. {}) to return all documents in the collection. projection (document) Optional. Controls the elds to return, or the projection. The projection argument will resemble the following prototype:
{ field1: boolean, field2: boolean ... }
The boolean can take the following include or exclude values: 1 or true to include. The find() (page 924) method always includes the _id eld even if the eld is not explicitly stated to return in the projection parameter. 0 or false to exclude.
924
The projection cannot contain both include and exclude specications except for the exclusion of the _id eld. Omit the projection parameter to return all the elds in the matching documents. See Projection Operators (page 796) for a list of special projection operators. Returns A cursor to the documents that match the query criteria. If the projection argument is specied, the matching documents contain only the projection elds, and the _id eld if you do not explicitly exclude the _id eld. Note: In the mongo (page 1036) shell, you can access the returned documents directly without explicitly using the JavaScript cursor handling method. Executing the query directly on the mongo (page 1036) shell prompt automatically iterates the cursor to display up to the rst 20 documents. Type it to continue iteration. Consider the following examples of the find() (page 924) method: To select all documents in a collection, call the find() (page 924) method with no parameters:
db.products.find()
This operation returns all the documents with all the elds from the collection products. By default, in the mongo (page 1036) shell, the cursor returns the rst batch of 20 matching documents. In the mongo (page 1036) shell, iterate through the next batch by typing it. Use the appropriate cursor handling mechanism for your specic language driver. To select the documents that match a selection criteria, call the find() (page 924) method with the query criteria:
db.products.find( { qty: { $gt: 25 } } )
This operation returns all the documents from the collection products where qty is greater than 25, including all elds. To select the documents that match a selection criteria and return, or project only certain elds into the result set, call the find() (page 924) method with the query criteria and the projection parameter, as in the following example:
db.products.find( { qty: { $gt: 25 } }, { item: 1, qty: 1 } )
This operation returns all the documents from the collection products where qty is greater than 25. The documents in the result set only include the _id, item, and qty elds using inclusion projection. find() (page 924) always returns the _id eld, even when not explicitly included:
{ "_id" : 11, "item" : "pencil", "qty" : 50 } { "_id" : ObjectId("50634d86be4617f17bb159cd"), "item" : "bottle", "qty" : 30 } { "_id" : ObjectId("50634dbcbe4617f17bb159d0"), "item" : "paper", "qty" : 100 }
To select the documents that match a query criteria and exclude a set of elds from the resulting documents, call the find() (page 924) method with the query criteria and the projection parameter using the exclude syntax:
db.products.find( { qty: { $gt: 25 } }, { _id: 0, qty: 0 } )
The query will return all the documents from the collection products where qty is greater than 25. The documents in the result set will contain all elds except the _id and qty elds, as in the following:
{ "item" : "pencil", "type" : "no.2" } { "item" : "bottle", "type" : "blue" } { "item" : "paper" }
925
db.collection.ndAndModify()
db.collection.findAndModify() The findAndModify() (page 926) method atomically modies and returns a single document. By default, the returned document does not include the modications made on the update. To return the document with the modications made on the update, use the new option. The findAndModify() (page 926) method is a shell helper around the findAndModify (page 826) command.
db.collection.findAndModify( { query: <document>, sort: <document>, remove: <boolean>, update: <document>, new: <boolean>, fields: <document>, upsert: <boolean> } );
The db.collection.findAndModify() (page 926) method takes a document parameter with the following subdocument elds: Fields query (document) Optional. Species the selection criteria for the modication. The query eld employs the same query selectors (page 763) as used in the db.collection.find() (page 924) method. Although the query may match multiple documents, findAndModify() (page 926) will only select one document to modify. sort (document) Optional. Determines which document the operation will modify if the query selects multiple documents. findAndModify() (page 926) will modify the rst document in the sort order specied by this argument. remove (boolean) Optional if update eld exists. When true, removes the selected document. The default is false. update (document) Optional if remove eld exists. Performs an update of the selected document. The update eld employs the same update operators (page 784) or field: value specications to modify the selected document. new (boolean) Optional. When true, returns the modied document rather than the original. The findAndModify() (page 926) method ignores the new option for remove operations. The default is false. elds (document) Optional. A subset of elds to return. The fields document species an inclusion of a eld with 1, as in the following:
fields: { <field1>: 1, <field2>: 1, ... }
See projection (page 171). upsert (boolean) Optional. Used in conjunction with the update eld. When true, findAndModify() (page 926) creates a new document if the query returns no documents. The default is false. The findAndModify() (page 926) method returns either: the pre-modication document or, if the new: Note: 926 Chapter 63. MongoDB Interface true option is set, the modied document.
If no document is found for the update or remove, the method returns null. If no document is found for an upsert, which means the command performs an insert, and new is false, and the sort option is NOT specied, the method returns null. Changed in version 2.2: Previously returned an empty document {}. See the 2.2 release notes (page 1160) for more information. If no document is found for an upsert, which means the command performs an insert, and new is false, and a sort option is specied, the method returns an empty document {}. Consider the following examples: The following method updates an existing document in the people collection where the document matches the query criteria:
db.people.findAndModify( { query: { name: "Tom", state: "active", rating: { $gt: 10 } }, sort: { rating: 1 }, update: { $inc: { score: 1 } } } )
This method performs the following actions: 1.The query nds a document in the people collection where the name eld has the value Tom, the state eld has the value active and the rating eld has a value greater than (page 764) 10. 2.The sort orders the results of the query in ascending order. If multiple documents meet the query condition, the method will select for modication the rst document as ordered by this sort. 3.The update increments (page 784) the value of the score eld by 1. 4.The method returns the original (i.e. pre-modication) document selected for this update:
{ "_id" : ObjectId("50f1e2c99beb36a0f45c6453"), "name" : "Tom", "state" : "active", "rating" : 100, "score" : 5 }
To return the modied document, add the new:true option to the method. If no document matched the query condition, the method returns null:
null
The following method includes the upsert: matches the query condition:
true option to insert a new document if no document
db.people.findAndModify( { query: { name: "Gus", state: "active", rating: 100 }, sort: { rating: 1 }, update: { $inc: { score: 1 } }, upsert: true } )
If the method does not nd a matching document, the method performs an upsert. Because the method included the sort option, it returns an empty document { } as the original (pre-modication) document:
927
{ }
If the method did not include a sort option, the method returns null.
null
The following method includes both the upsert: true option and the new:true option to return the newly inserted document if a document matching the query is not found:
db.people.findAndModify( { query: { name: "Pascal", state: "active", rating: 25 }, sort: { rating: 1 }, update: { $inc: { score: 1 } }, upsert: true, new: true } )
The method returns the newly inserted document:

{ "_id" : ObjectId("50f49ad6444c11ac2448a5d6"), "name" : "Pascal", "rating" : 25, "score" : 1, "state" : "active" }
When findAndModify() (page 926) includes the upsert: true option and the query eld(s) is not uniquely indexed, the method could insert a document multiple times in certain circumstances. For instance, if multiple clients each invoke the method with the same query condition and these methods complete the find phase before any of methods perform the modify phase, these methods could insert the same document. Consider an example where no document with the name Andy exists and multiple clients issue the following command:
db.people.findAndModify( { query: { name: "Andy" }, sort: { rating: 1 }, update: { $inc: { score: 1 } }, upsert: true } )
If all the methods nish the query phase before any command starts the modify phase, and there is no unique index on the name eld, the commands may all perform an upsert. To prevent this condition, create a unique index (page 334) on the name eld. With the unique index in place, the multiple methods would observe one of the following behaviors: Exactly one findAndModify() (page 926) would successfully insert a new document. Zero or more findAndModify() (page 926) methods would update the newly inserted document. Zero or more findAndModify() (page 926) methods would fail when they attempted to insert a duplicate. If the method fails due to a unique index constraint violation, you can retry the method. Absent a delete of the document, the retry should not fail.
928
Warning: When using findAndModify (page 826) in a sharded environment, the query must contain the shard key for all operations against the shard cluster. findAndModify (page 826) operations issued against mongos (page 1032) instances for non-sharded collections function normally.
db.collection.ndOne()
db.collection.findOne(query, projection) Parameters query (document) Optional. A document that species the query using the JSON-like syntax and query operators (page 763). projection (document) Optional. Controls the elds to return, or the projection. The projection argument will resemble the following prototype:
{ field1: boolean, field2: boolean ... }
The boolean can take the following include or exclude values: 1 or true to include. The findOne() (page 929) method always includes the _id eld even if the eld is not explicitly stated to return in the projection parameter. 0 or false to exclude. The projection cannot contain both include and exclude specications except for the exclusion of the _id eld. Omit the projection parameter to return all the elds in the matching documents. Returns One document that satises the query specied as the argument to this method. If the projection argument is specied, the returned document contains only the projection elds, and the _id eld if you do not explicitly exclude the _id eld. Returns only one document that satises the specied query. If multiple documents satisfy the query, this method returns the rst document according to the natural order which reects the order of documents on the disc. In capped collections, natural order is the same as insertion order.
db.collection.getIndexStats()
Denition db.collection.getIndexStats({ index: <index name> }) Parameters index (document) The index name (page 335). The getIndexStats (page 929) method displays a human-readable summary of aggregated statistics about an indexs B-tree data structure. The information summarizes the output returned by the indexStats (page 883) command and db.collection.indexStats() method. The getIndexStats (page 929) method displays the information on the screen and does not return an object. db.collection.getIndexStats() (page 929) is only available when connected to a mongod (page 1021) instance that uses the --enableExperimentalIndexStatsCmd option. Consider the following prototype invocation:
929
db.<collection>.getIndexStats( { index : "<index name>" } )
To view index names (page 335) for a collection, use the getIndexes() (page 930) method. Warning: Do not use db.collection.getIndexStats() (page 929) or indexStats (page 883) with production deployments.
Example The following command returns information for an index named type_1_traits_1:
db.animals.getIndexStats({index:"type_1_traits_1"})
The command returns the following summary. For more information on the B-tree statistics, see indexStats (page 883).
-- index "undefined" -version 1 | key pattern { "type" : 1, 2 deep, bucket body is 8154 bytes bucket count -- depth 0 -bucket count -- depth 1 -bucket count -- depth 2 -bucket count 45513
"traits" : 1 } | storage namespace "test.animals.$type_1_tr
on average 99.401 % (0.463 %) full
49.581 % (4.135 %) bson keys,
on average 71.511 % (0.000 %) full
36.191 % (0.000 %) bson keys,
180
on average 98.954 % (5.874 %) full
49.732 % (5.072 %) bson keys,
45332
on average 99.403 % (0.245 %) full
49.580 % (4.130 %) bson keys,
db.collection.getIndexes()
db.collection.getIndexes() Returns an array that holds a list of documents that identify and describe the existing indexes on the collection. You must call the db.collection.getIndexes() (page 930) on a collection. For example:
db.collection.getIndexes()
Change collection to the name of the collection whose indexes you want to learn. The db.collection.getIndexes() (page 930) items consist of the following elds: system.indexes.v Holds the version of the index. The index version depends on the version of mongod (page 1021) that created the index. Before version 2.0 of MongoDB, the this value was 0; versions 2.0 and later use version 1. system.indexes.key Contains a document holding the keys held in the index, and the order of the index. Indexes may be either descending or ascending order. A value of negative one (e.g. -1) indicates an index sorted in descending order while a positive value (e.g. 1) indicates an index sorted in an ascending order. system.indexes.ns The namespace context for the index. system.indexes.name A unique name for the index comprised of the eld names and orders of all keys. 930 Chapter 63. MongoDB Interface
db.collection.getShardDistribution()
db.collection.getShardDistribution() Returns Prints the data distribution statistics for a sharded collection. You must call the getShardDistribution() (page 931) method on a sharded collection, as in the following example:
db.myShardedCollection.getShardDistribution()
In the following example, the collection has two shards. The output displays both the individual shard distribution information as well the total shard distribution:
Shard <shard-a> at <host-a> data : <size-a> docs : <count-a> chunks : <number of chunks-a> estimated data per chunk : <size-a>/<number of chunks-a> estimated docs per chunk : <count-a>/<number of chunks-a> Shard <shard-b> at <host-b> data : <size-b> docs : <count-b> chunks : <number of chunks-b> estimated data per chunk : <size-b>/<number of chunks-b> estimated docs per chunk : <count-b>/<number of chunks-b>
Totals data : <stats.size> docs : <stats.count> chunks : <calc total chunks> Shard <shard-a> contains <estDataPercent-a>% data, <estDocPercent-a>% docs in cluster, avg obj Shard <shard-b> contains <estDataPercent-b>% data, <estDocPercent-b>% docs in cluster, avg obj
The output information displays: <shard-x> is a string that holds the shard name. <host-x> is a string that holds the host name(s). <size-x> is a number that includes the size of the data, including the unit of measure (e.g. b, Mb). <count-x> is a number that reports the number of documents in the shard. <number of chunks-x> is a number that reports the number of chunks in the shard. <size-x>/<number of chunks-x> is a calculated value that reects the estimated data size per chunk for the shard, including the unit of measure (e.g. b, Mb). <count-x>/<number of chunks-x> is a calculated value that reects the estimated number of documents per chunk for the shard. <stats.size> is a value that reports the total size of the data in the sharded collection, including the unit of measure. <stats.count> is a value that reports the total number of documents in the sharded collection. <calc total chunks> is a calculated number that reports the number of chunks from all shards, for example:
<calc total chunks> = <number of chunks-a> + <number of chunks-b>
<estDataPercent-x> is a calculated value that reects, for each shard, the data size as the percentage of the collections total data size, for example:
931
<estDataPercent-x> = <size-x>/<stats.size>
<estDocPercent-x> is a calculated value that reects, for each shard, the number of documents as the percentage of the total number of documents for the collection, for example:
<estDocPercent-x> = <count-x>/<stats.count>
stats.shards[ <shard-x> ].avgObjSize is a number that reects the average object size, including the unit of measure, for the shard. For example, the following is a sample output for the distribution of a sharded collection:
Shard shard-a at shard-a/MyMachine.local:30000,MyMachine.local:30001,MyMachine.local:30002 data : 38.14Mb docs : 1000003 chunks : 2 estimated data per chunk : 19.07Mb estimated docs per chunk : 500001 Shard shard-b at shard-b/MyMachine.local:30100,MyMachine.local:30101,MyMachine.local:30102 data : 38.14Mb docs : 999999 chunks : 3 estimated data per chunk : 12.71Mb estimated docs per chunk : 333333 Totals data : 76.29Mb docs : 2000002 chunks : 5 Shard shard-a contains 50% data, 50% docs in cluster, avg obj size on shard : 40b Shard shard-b contains 49.99% data, 49.99% docs in cluster, avg obj size on shard : 40b
See also: Sharding (page 483)

db.collection.getShardVersion()
db.collection.getShardVersion() This method returns information regarding the state of data in a sharded cluster that is useful when diagnosing underlying issues with a sharded cluster. For internal and diagnostic use only.
db.collection.group()
db.collection.group({ key, reduce, initial, [keyf,] [cond,] nalize }) The db.collection.group() (page 932) method groups documents in a collection by the specied keys and performs simple aggregation functions such as computing counts and sums. The method is analogous to a SELECT .. GROUP BY statement in SQL. The group() (page 932) method returns an array. The db.collection.group() (page 932) accepts a single document that contains the following: Fields key Species one or more document elds to group by. reduce Species a function for the group operation perform on the documents during the grouping operation, such as compute a sum or a count. The aggregation function takes two arguments: the current document and the aggregate result for the previous documents in the. initial Initializes the aggregation result document.
932
keyf Optional. Alternative to the key eld. Species a function that creates a key object for use as the grouping key. Use the keyf instead of key to group by calculated elds rather than existing document elds. cond Optional. Species the selection criteria to determine which documents in the collection to process. If you omit the cond eld, db.collection.group() (page 932) processes all the documents in the collection for the group operation. nalize Optional. Species a function that runs each item in the result set before db.collection.group() (page 932) returns the nal value. This function can either modify the result document or replace the result document as a whole. The db.collection.group() (page 932) method is a shell wrapper for the group (page 810) command; however, the db.collection.group() (page 932) method takes the keyf eld and the reduce eld whereas the group (page 810) command takes the $keyf eld and the $reduce eld. Warning: The db.collection.group() (page 932) method does not work with sharded clusters. Use the aggregation framework or map-reduce in sharded environments. The result set must t within the maximum BSON document size (page 1105). In version 2.2, the returned array can contain at most 20,000 elements; i.e. at most 20,000 unique groupings. For group by operations that results in more than 20,000 unique groupings, use mapReduce (page 814). Previous versions had a limit of 10,000 elements. Prior to 2.4, the db.collection.group() (page 932) method took the mongod (page 1021) instances JavaScript lock, which blocked all other JavaScript execution.
Note: Changed in version 2.4. In MongoDB 2.4, map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions cannot access certain global functions or properties, such as db, that are available in the mongo (page 1036) shell. When upgrading to MongoDB 2.4, you will need to refactor your code if your map-reduce operations (page 814), group (page 810) commands, or $where (page 775) operator expressions include any global shell functions or properties that are no longer available, such as db. The following shell functions and properties are available to map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions in MongoDB 2.4:
933
Consider the following examples of the db.collection.group() (page 932) method: The examples assume an orders collection with documents of the following prototype:
{ _id: ObjectId("5085a95c8fada716c89d0021"), ord_dt: ISODate("2012-07-01T04:00:00Z"), ship_dt: ISODate("2012-07-02T04:00:00Z"), item: { sku: "abc123", price: 1.99, uom: "pcs", qty: 25 } }
The following example groups by the ord_dt and item.sku elds those documents that have ord_dt greater than 01/01/2011:
db.orders.group( { key: { ord_dt: 1, item.sku: 1 }, cond: { ord_dt: { $gt: new Date( 01/01/2012 ) } }, reduce: function ( curr, result ) { }, initial: { } } )
The result is an array of documents that contain the group by elds:

[ { { { { { { { { { { "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" : : : : : : : : : : ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-05-01T04:00:00Z"), ISODate("2012-05-01T04:00:00Z"), ISODate("2012-06-08T04:00:00Z"), "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" : : : : : : : : : : "abc123"}, "abc456"}, "bcd123"}, "efg456"}, "abc123"}, "efg456"}, "ijk123"}, "abc123"}, "abc456"}, "abc123"},
934
{ "ord_dt" : ISODate("2012-06-08T04:00:00Z"), "item.sku" : "abc456"} ]

SELECT ord_dt, item_sku FROM orders WHERE ord_dt > 01/01/2012 GROUP BY ord_dt, item_sku
The following example groups by the ord_dt and item.sku elds, those documents that have ord_dt greater than 01/01/2011 and calculates the sum of the qty eld for each grouping:
db.orders.group( { key: { ord_dt: 1, item.sku: 1 }, cond: { ord_dt: { $gt: new Date( 01/01/2012 ) } }, reduce: function ( curr, result ) { result.total += curr.item.qty; }, initial: { total : 0 } } )
The result is an array of documents that contain the group by elds and the calculated aggregation eld:
[ { { { { { { { { { { { "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" "ord_dt" : : : : : : : : : : : ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-07-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-06-01T04:00:00Z"), ISODate("2012-05-01T04:00:00Z"), ISODate("2012-05-01T04:00:00Z"), ISODate("2012-06-08T04:00:00Z"), ISODate("2012-06-08T04:00:00Z"), "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" "item.sku" : : : : : : : : : : : "abc123", "abc456", "bcd123", "efg456", "abc123", "efg456", "ijk123", "abc123", "abc456", "abc123", "abc456", "total" "total" "total" "total" "total" "total" "total" "total" "total" "total" "total" : : : : : : : : : : : 25 25 10 10 25 15 20 45 25 25 25 }, }, }, }, }, }, }, }, }, }, } ]

SELECT ord_dt, item_sku, SUM(item_qty) as total FROM orders WHERE ord_dt > 01/01/2012 GROUP BY ord_dt, item_sku
The following example groups by the calculated day_of_week eld, those documents that have ord_dt greater than 01/01/2011 and calculates the sum, count, and average of the qty eld for each grouping:
db.orders.group( { keyf: function(doc) { return { day_of_week: doc.ord_dt.getDay() } ; }, cond: { ord_dt: { $gt: new Date( 01/01/2012 ) } }, reduce: function ( curr, result ) { result.total += curr.item.qty; result.count++; }, initial: { total : 0, count: 0 }, finalize: function(result) { var weekdays = [ "Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday" ]; result.day_of_week = weekdays[result.day_of_week];
935
result.avg = Math.round(result.total / result.count); } } )
The result is an array of documents that contain the group by elds and the calculated aggregation eld:
[ { "day_of_week" : "Sunday", "total" : 70, "count" : 4, "avg" : 18 }, { "day_of_week" : "Friday", "total" : 110, "count" : 6, "avg" : 18 }, { "day_of_week" : "Tuesday", "total" : 70, "count" : 3, "avg" : 23 } ]
See also: Aggregation Framework (page 255)

db.collection.insert()
db.collection.insert(document) The insert() (page 936) method inserts a document or documents into a collection. Changed in version 2.2: The insert() (page 936) method can accept an array of documents to perform a bulk insert of the documents into the collection. Consider the following behaviors of the insert() (page 936) method: If the collection does not exist, then the insert() (page 936) method will create the collection. If the document does not specify an _id eld, then MongoDB will add the _id eld and assign a unique ObjectId for the document before inserting. Most drivers create an ObjectId and insert the _id eld, but the mongod (page 1021) will create and populate the _id if the driver or application does not. If the document species a new eld, then the insert() (page 936) method inserts the document with the new eld. This requires no changes to the data model for the collection or the existing documents. The insert() (page 936) method takes one of the following parameters: Parameters document A document to insert into the collection. documents (array) New in version 2.2. An array of documents to insert into the collection. Consider the following examples of the insert() (page 936) method: To insert a single document and have MongoDB generate the unique _id, omit the _id eld in the document and pass the document to the insert() (page 936) method as in the following:
db.products.insert( { item: "card", qty: 15 } )
This operation inserts a new document into the products collection with the item eld set to card, the qty eld set to 15, and the _id eld set to a unique ObjectId:
{ "_id" : ObjectId("5063114bd386d8fadbd6b004"), "item" : "card", "qty" : 15 }
Note: Most MongoDB driver clients will include the _id eld and generate an ObjectId before sending the insert operation to MongoDB; however, if the client sends a document without an _id eld, the mongod (page 1021) will add the _id eld and generate the ObjectId.
936
To insert a single document, with a custom _id eld, include the _id eld set to a unique identier and pass the document to the insert() (page 936) method as follows:
db.products.insert( { _id: 10, item: "box", qty: 20 } )
This operation inserts a new document in the products collection with the _id eld set to 10, the item eld set to box, the qty eld set to 20:
{ "_id" : 10, "item" : "box", "qty" : 20 }
Note: Most MongoDB driver clients will include the _id eld and generate an ObjectId before sending the insert operation to MongoDB; however, if the client sends a document without an _id eld, the mongod (page 1021) will add the _id eld and generate the ObjectId. To insert multiple documents, pass an array of documents to the insert() (page 936) method as in the following:
db.products.insert( [ { _id: 11, item: "pencil", qty: 50, type: "no.2" }, { item: "pen", qty: 20 }, { item: "eraser", qty: 25 } ] )
The operation will insert three documents into the products collection: A document with the elds _id set to 11, item set to pencil, qty set to 50, and the type set to no.2. A document with the elds _id set to a unique objectid, item set to pen, and qty set to 20. A document with the elds _id set to a unique objectid, item set to eraser, and qty set to 25.
{ "_id" : 11, "item" : "pencil", "qty" : 50, "type" : "no.2" } { "_id" : ObjectId("50631bc0be4617f17bb159ca"), "item" : "pen", "qty" : 20 } { "_id" : ObjectId("50631bc0be4617f17bb159cb"), "item" : "eraser", "qty" : 25 }
db.collection.isCapped()
db.collection.isCapped() Returns Returns true if the collection is a capped collection, otherwise returns false. See also: Capped Collections (page 558)
db.collection.mapReduce()
db.collection.mapReduce(map, reduce, {<out>, <query>, <sort>, <limit>, <nalize>, <scope>, <jsMode>, <verbose>}) The db.collection.mapReduce() (page 937) method provides a wrapper around the mapReduce (page 814) command.
db.collection.mapReduce( <map>, <reduce>, { out: <collection>, query: <document>,
937
sort: <document>, limit: <number>, finalize: <function>, scope: <document>, jsMode: <boolean>, verbose: <boolean> } )
db.collection.mapReduce() (page 937) takes the following parameters: Parameters map A JavaScript function that associates or maps a value with a key and emits the key and value pair. The map function processes every input document for the map-reduce operation. The mapreduce operation groups the emitted value objects by the key and passes these groupings to the reduce function. reduce A JavaScript function that reduces to a single object all the values associated with a particular key. The reduce function accepts two arguments: key and values. The values argument is an array whose elements are the value objects that are mapped to the key. out New in version 1.8. Species the location of the result of the map-reduce operation. You can output to a collection, output to a collection with an action, or output inline. You may output to a collection when performing map reduce operations on the primary members of the set; on secondary members you may only use the inline output. query Optional. Species the selection criteria using query operators (page 763) for determining the documents input to the map function. sort Optional. Sorts the input documents. This option is useful for optimization. For example, specify the sort key to be the same as the emit key so that there are fewer reduce operations. limit Optional. Species a maximum number of documents to return from the collection. nalize Optional. A JavaScript function that follows the reduce method and modies the output. The finalize function receives two arguments: key and reducedValue. reducedValue is the value returned from the reduce function for the key. The
scope (document) Optional. Species global variables that are accessible in the map , reduce and the finalize functions. jsMode (Boolean) New in version 2.0. Optional. Species whether to convert intermediate data into BSON format between the execution of the map and reduce functions. If false: Internally, MongoDB converts the JavaScript objects emitted by the map function to BSON objects. These BSON objects are then converted back to JavaScript objects when calling the reduce function.
938
The map-reduce operation places the intermediate BSON objects in temporary, on-disk storage. This allows the map-reduce operation to execute over arbitrarily large data sets. If true: Internally, the JavaScript objects emitted during map function remain as JavaScript objects. There is no need to convert the objects for the reduce function, which can result in faster execution. You can only use jsMode for result sets with fewer than 500,000 distinct key arguments to the mappers emit() function. The jsMode defaults to false. verbose (Boolean) Optional. Species whether to include the timing information in the result information. The verbose defaults to true to include the timing information. Note: Changed in version 2.4. In MongoDB 2.4, map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions cannot access certain global functions or properties, such as db, that are available in the mongo (page 1036) shell. When upgrading to MongoDB 2.4, you will need to refactor your code if your map-reduce operations (page 814), group (page 810) commands, or $where (page 775) operator expressions include any global shell functions or properties that are no longer available, such as db. The following shell functions and properties are available to map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions in MongoDB 2.4: Available Properties args MaxKey MinKey Available Functions assert() BinData() DBPointer() DBRef() doassert() emit() gc() HexData() hex_md5() isNumber() isObject() ISODate() isString() Map() MD5() NumberInt() NumberLong() ObjectId() print() printjson() printjsononeline() sleep() Timestamp() tojson() tojsononeline() tojsonObject() UUID() version()
Requirements for the map Function The map function has the following prototype:
function() { ... emit(key, value); }
939
The map function exhibits the following behaviors: In the map function, reference the current document as this within the function. The map function should not access the database for any reason. The map function should be pure, or have no impact outside of the function (i.e. side effects.) The emit(key,value) function associates the key with a value. A single emit can only hold half of MongoDBs maximum BSON document size (page 1105). There is no limit to the number of times you may call the emit function per document. The map function can access the variables dened in the scope parameter. Requirements for the reduce Function The reduce function has the following prototype:
function(key, values) { ... return result; }
The reduce function exhibits the following behaviors: The reduce function should not access the database, even to perform read operations. The reduce function should not affect the outside system. MongoDB will not call the reduce function for a key that has only a single value. The reduce function can access the variables dened in the scope parameter. Because it is possible to invoke the reduce function more than once for the same key, the following properties need to be true: the type of the return object must be identical to the type of the value emitted by the map function to ensure that the following operations is true:
reduce(key, [ C, reduce(key, [ A, B ]) ] ) == reduce( key, [ C, A, B ] )
the reduce function must be idempotent. Ensure that the following statement is true:
reduce( key, [ reduce(key, valuesArray) ] ) == reduce( key, valuesArray )
the order of the elements in the valuesArray should not affect the output of the reduce function, so that the following statement is true:
reduce( key, [ A, B ] ) == reduce( key, [ B, A ] )
out Options You can specify the following options for the out parameter: Output to a Collection
out: <collectionName>
Output to a Collection with an Action This option is only available when passing out a collection that already exists. This option is not available on secondary members of replica sets.
940
out: { <action>: <collectionName> [, db: <dbName>] [, sharded: <boolean> ] [, nonAtomic: <boolean> ] }
When you output to a collection with an action, the out has the following parameters: <action>: Specify one of the following actions: replace Replace the contents of the <collectionName> if the collection with the <collectionName> exists. merge Merge the new result with the existing result if the output collection already exists. If an existing document has the same key as the new result, overwrite that existing document. reduce Merge the new result with the existing result if the output collection already exists. If an existing document has the same key as the new result, apply the reduce function to both the new and the existing documents and overwrite the existing document with the result. db: Optional.The name of the database that you want the map-reduce operation to write its output. By default this will be the same database as the input collection. sharded: Optional. If true and you have enabled sharding on output database, the map-reduce operation will shard the output collection using the _id eld as the shard key. nonAtomic: New in version 2.2. Optional. Specify output operation as non-atomic and is valid only for merge and reduce output modes which may take minutes to execute. If nonAtomic is true, the post-processing step will prevent MongoDB from locking the database; however, other clients will be able to read intermediate states of the output collection. Otherwise the map reduce operation must lock the database during post-processing. Output Inline Perform the map-reduce operation in memory and return the result. This option is the only available option for out on secondary members of replica sets.
out: { inline: 1 }
The result must t within the maximum size of a BSON document (page 1105). Requirements for the finalize Function The finalize function has the following prototype:
function(key, reducedValue) { ... return modifiedObject; }
The finalize function receives as its arguments a key value and the reducedValue from the reduce function. Be aware that: 63.3. mongo Shell Methods 941
The finalize function should not access the database for any reason. The finalize function should be pure, or have no impact outside of the function (i.e. side effects.) The finalize function can access the variables dened in the scope parameter. Map-Reduce Examples Consider the following map-reduce operations on a collection orders that contains documents of the following prototype:
1. Dene the map function to process each input document: In the function, this refers to the document that the map-reduce operation is processing. The function maps the price to the cust_id for each document and emits the cust_id and price pair.
This operation outputs the results to a collection named map_reduce_example. If the map_reduce_example collection already exists, the operation will replace the contents with the results of this map-reduce operation:
942
Calculate Order and Total Quantity with Average Quantity Per Item In this example you will perform a map-reduce operation on the orders collection, for all documents that have an ord_date value greater than 01/01/2012. The operation groups by the item.sku eld, and for each sku calculates the number of orders and the total quantity ordered. The operation concludes by calculating the average quantity per order for each sku value: 1. Dene the map function to process each input document: In the function, this refers to the document that the map-reduce operation is processing. For each item, the function associates the sku with a new object value that contains the count of 1 and the item qty for the order and emits the sku and value pair.
2. Dene the corresponding reduce function with two arguments keySKU and countObjVals: countObjVals is an array whose elements are the objects mapped to the grouped keySKU values passed by map function to the reducer function. The function reduces the countObjVals array to a single object reducedValue that contains the count and the qty elds. In reducedVal, the count eld contains the sum of the count elds from the individual array elements, and the qty eld contains the sum of the qty elds from the individual array elements.
using
the
mapFunction2,
943
This operation uses the query eld to select only those documents with ord_date greater than new Date(01/01/2012). Then it output the results to a collection map_reduce_example. If the map_reduce_example collection already exists, the operation will merge the existing contents with the results of this map-reduce operation: For more information and examples, see the Map-Reduce (page 311) page and Perform Incremental Map-Reduce (page 313). See also: Troubleshoot the Map Function (page 317) Troubleshoot the Reduce Function (page 318) mapReduce (page 814) command Aggregation Framework (page 255)
db.collection.reIndex()
db.collection.reIndex() This method drops all indexes and recreates them. This operation may be expensive for collections that have a large amount of data and/or a large number of indexes. Call this method, which takes no arguments, on a collection object. For example:
db.collection.reIndex()
Change collection to the name of the collection that you want to rebuild the index.
db.collection.remove()
db.collection.remove(query, justOne) The remove (page 944) method removes documents from a collection. The remove() (page 944) method can take the following parameters: Parameters query (document) Optional. Species the deletion criteria using query operators (page 763). Omit the query parameter or pass an empty document (e.g. {} ) to delete all documents in the collection. justOne (boolean) Optional. A boolean that limits the deletion to just one document. The default value is false. Set to true to delete only the rst result. Note: You cannot use the remove() (page 944) method with a capped collection.
944
Consider the following examples of the remove (page 944) method. To remove all documents in a collection, call the remove (page 944) method with no parameters:
db.products.remove()
This operation will remove all the documents from the collection products. To remove the documents that match a deletion criteria, call the remove (page 944) method with the query criteria:
db.products.remove( { qty: { $gt: 20 } } )
This operation removes all the documents from the collection products where qty is greater than 20. To remove the rst document that match a deletion criteria, call the remove (page 944) method with the query criteria and the justOne parameter set to true or 1:
db.products.remove( { qty: { $gt: 20 } }, true )
This operation removes the rst document from the collection products where qty is greater than 20. Note: If the query argument to the remove() (page 944) method matches multiple documents in the collection, the delete operation may interleave with other write operations to that collection. For an unsharded collection, you have the option to override this behavior with the $isolated (page 796) isolation operator, effectively isolating the delete operation and blocking other write operations during the delete. To isolate the query, include $isolated: 1 in the query parameter as in the following example:
db.products.remove( { qty: { $gt: 20 }, $isolated: 1 } )
db.collection.renameCollection()
db.collection.renameCollection() db.collection.renameCollection() (page 945) provides a helper for the renameCollection (page 867) database command in the mongo (page 1036) shell to rename existing collections. Parameters target (string) Species the new name of the collection. Enclose the string in quotes. dropTarget (boolean) Optional. If true, mongod (page 1021) will drop the target of renameCollection (page 867) prior to renaming the collection. Call the db.collection.renameCollection() (page 945) method on a collection object, to rename a collection. Specify the new name of the collection as an argument. For example:
db.rrecord.renameCollection("record")
This operation will rename the rrecord collection to record. If the target name (i.e. record) is the name of an existing collection, then the operation will fail. Consider the following limitations: db.collection.renameCollection() (page 945) cannot move a collection between databases. Use renameCollection (page 867) for these rename operations. db.collection.renameCollection() (page 945) cannot operation on sharded collections. The db.collection.renameCollection() (page 945) method operates within a collection by changing the metadata associated with a given collection.
945
Refer to the documentation renameCollection (page 867) for additional warnings and messages. Warning: The db.collection.renameCollection() (page 945) method and renameCollection (page 867) command will invalidate open cursors which interrupts queries that are currently returning data.
db.collection.save()
db.collection.save(document) The save() (page 946) method updates an existing document or inserts a document depending on the parameter. The save() (page 946) method takes the following parameter: Parameters document Specify a document to save to the collection. If the document does not contain an _id eld, then the save() (page 946) method performs an insert with the specied elds in the document as well as an _id eld with a unique objectid value. If the document contains an _id eld, then the save() (page 946) method performs an upsert querying the collection on the _id eld: If a document does not exist with the specied _id value, the save() (page 946) method performs an insert with the specied elds in the document. If a document exists with the specied _id value, the save() (page 946) method performs an update, replacing all eld in the existing record with the elds from the document. Consider the following examples of the save() (page 946) method: Pass to the save() (page 946) method a document without an _id eld, so that to insert the document into the collection and have MongoDB generate the unique _id as in the following:
db.products.save( { item: "book", qty: 40 } )
This operation inserts a new document into the products collection with the item eld set to book, the qty eld set to 40, and the _id eld set to a unique ObjectId:
{ "_id" : ObjectId("50691737d386d8fadbd6b01d"), "item" : "book", "qty" : 40 }
Note: Most MongoDB driver clients will include the _id eld and generate an ObjectId before sending the insert operation to MongoDB; however, if the client sends a document without an _id eld, the mongod (page 1021) will add the _id eld and generate the ObjectId. Pass to the save() (page 946) method a document with an _id eld that holds a value that does not exist in the collection to insert the document with that value in the _id value into the collection, as in the following:
db.products.save( { _id: 100, item: "water", qty: 30 } )
This operation inserts a new document into the products collection with the _id eld set to 100, the item eld set to water, and the eld qty set to 30:
946
{ "_id" : 100, "item" : "water", "qty" : 30 }
Note: Most MongoDB driver clients will include the _id eld and generate an ObjectId before sending the insert operation to MongoDB; however, if the client sends a document without an _id eld, the mongod (page 1021) will add the _id eld and generate the ObjectId. Pass to the save() (page 946) method a document with the _id eld set to a value in the collection to replace all elds and values of the matching document with the new elds and values, as in the following:
db.products.save( { _id:100, item:"juice" } )
This operation replaces the existing document with a value of 100 in the _id eld. The updated document will resemble the following:
{ "_id" : 100, "item" : "juice" }
db.collection.stats()
db.collection.stats(scale) Parameters scale Optional. Species the scale to deliver results. Unless specied, this command returns all sizes in bytes. Returns A document containing statistics that reecting the state of the specied collection. This function provides a wrapper around the database command collStats (page 873). The scale option allows you to congure how the mongo (page 1036) shell scales the sizes of things in the output. For example, specify a scale value of 1024 to display kilobytes rather than bytes. Call the db.collection.stats() (page 947) method on a collection object, to return statistics regarding that collection. For example, the following operation returns stats on the people collection:
db.people.stats()
See also: collStats (page 873) for an overview of the output of this command.
db.collection.storageSize()
db.collection.storageSize() Returns The total amount of storage allocated to this collection for document storage. Provides a wrapper around the storageSize (page 874) eld of the collStats (page 873) (i.e. db.collection.stats() (page 947)) output.
db.collection.storageSize()
db.collection.storageSize() Returns The total amount of storage allocated to this collection for document storage. Provides a wrapper around the storageSize (page 874) eld of the collStats (page 873) (i.e. db.collection.stats() (page 947)) output.
947
db.collection.totalSize()
db.collection.totalSize() Returns The total size of the data in the collection plus the size of every indexes on the collection.
db.collection.update()
db.collection.update(query, update[, options ]) The update() (page 948) method modies an existing document or documents in a collection. By default the update() (page 948) method updates a single document. To update all documents in the collection that match the update query criteria, specify the multi option. To insert a document if no document matches the update query criteria, specify the upsert option. Changed in version 2.2: The mongo (page 1036) shell provides an updated interface that accepts the options parameter in a document format to specify multi and upsert options. Prior to version 2.2, in the mongo (page 1036) shell, upsert and multi were positional boolean options:
db.collection.update(query, update, <upsert>, <multi>)
The update() (page 948) method takes the following parameters: Parameters query (document) Species the selection criteria for the update. The query parameter employs the same query selectors (page 763) as used in the db.collection.find() (page 924) method. update (document) Species the modications to apply. If the update parameter contains any update operators (page 784) expressions such as the $set (page 788) operator expression, then: the update parameter must contain only update operators expressions. the update() (page 948) method updates only the corresponding elds in the document. If the update parameter consists only of field: value expressions, then:
the update() (page 948) method replaces the document with the updates document. If the updates document is missing the _id eld, MongoDB will add the _id eld and assign to it a unique objectid . the update() (page 948) method updates cannot update multiple documents. options (document) New in version 2.2. Optional. Species whether to perform an upsert and/or a multiple update. Use the options parameter instead of the individual upsert and multi parameters. upsert (boolean) Optional. Species an upsert operation The default value is false. When true, the update() (page 948) method will update an existing document that matches the query selection criteria or if no document matches the criteria, insert a new document with the elds and values of the update parameter and if the update included only update operators, the query parameter as well . In version 2.2 of the mongo (page 1036) shell, you may also specify upsert in the options parameter. Note: With upsert update() (page 948) inserts a single document. 948 Chapter 63. MongoDB Interface
multi (boolean) Optional. Species whether to update multiple documents that meet the query criteria. When not specied, the default value is false and the update() (page 948) method updates a single document that meet the query criteria. When true, the update() (page 948) method updates all documents that meet the query criteria. In version 2.2 of the mongo (page 1036) shell, you may also specify multi in the options parameter. Note: The multi update operation may interleave with other write operations. For unsharded collections, you can override this behavior with the $isolated (page 796) isolation operator, which isolates the update operation and blocks other write operations during the update. See the isolation operator (page 796). Although the update operation may apply mostly to updating the values of the elds, the update() (page 948) method can also modify the name of the field in a document using the $rename (page 784) operator. Consider the following examples of the update() (page 948) method. These examples all use the 2.2 interface to specify options in the document form. To update specic elds in a document, call the update() (page 948) method with an update parameter using field: value pairs and expressions using update operators (page 784) as in the following:
db.products.update( { item: "book", qty: { $gt: 5 } }, { $set: { x: 6 }, $inc: { y: 5} } )
This operation updates a document in the products collection that matches the query criteria and sets the value of the eld x to 6, and increment the value of the eld y by 5. All other elds of the document remain the same. To replace all the elds in a document with the document as specied in the update parameter, call the update() (page 948) method with an update parameter that consists of only key: value expressions, as in the following:
db.products.update( { item: "book", qty: { $gt: 5 } }, { x: 6, y: 15 } )
This operation selects a document from the products collection that matches the query criteria sets the value of the eld x to 6 and the value of the eld y to 15. All other elds of the matched document are removed, except the _id eld. To update multiple documents, call the update() (page 948) method and specify the multi option in the options argument, as in the following:
db.products.update( { item: "book", qty: { $gt: 5 } }, { $set: { x: 6, y: 15 } }, { multi: t
This operation updates all documents in the products collection that match the query criteria by setting the value of the eld x to 6 and the value of the eld y to 15. This operation does not affect any other elds in documents in the products collection. You can perform the same operation by calling the update() (page 948) method with the multi parameter:
db.products.update( { item: "book", qty: { $gt: 5 } }, { $set: { x: 6, y: 15 } }, false, tru
To update a document or to insert a new document if no document matches the query criteria, call the update() (page 948) and specify the upsert option in the options argument, as in the following:
949
db.products.update( { item: "magazine", qty: { $gt: 5 } }, { $set: { x: 25, y: 50 } }, { ups
This operation will: update a single document in the products collection that matches the query criteria, setting the value of the eld x to 25 and the value of the eld y to 50, or if no matching document exists, insert a document in the products collection, with the eld item set to magazine, the eld x set to 25, and the eld y set to 50.
db.collection.validate()
db.collection.validate() Parameters full (Boolean) Optional. Specify true to enable a full validation. MongoDB disables full validation by default because it is a potentially resource intensive operation. Provides a wrapper around the validate (page 907) database command. Call the db.collection.validate() (page 950) method on a collection object, to validate the collection itself. Specify the full option to return full statistics. The validation (page 907) operation scans all of the data structures for correctness and returns a single document that describes the relationship between the logical collection and the physical representation of that data. The output can provide a more in depth view of how the collection uses storage. Be aware that this command is potentially resource intensive, and may impact the performance of your MongoDB instance. See also: validate (page 907)
950
63.3.2 Cursor
Cursor Methods Name Description cursor.addOption() Adds special wire protocol ags that modify the behavior of the query. (page 951) cursor.batchSize() Controls the number of documents MongoDB will return to the client in a single (page 952) network message. cursor.count() Returns a count of the documents in a cursor. (page 952) cursor.explain() Reports on the query execution plan, including index use, for a cursor. (page 953) cursor.forEach() Applies a JavaScript function for every document in a cursor. (page 958) cursor.hasNext() Returns true if the cursor has documents and can be iterated. (page 958) cursor.hint() Forces MongoDB to use a specic index for a query. (page 958) cursor.limit() Constrains the size of a cursors result set. (page 959) cursor.map() Applies a function to each document in a cursor and collects the return values in an (page 959) array. cursor.max() Species an exclusive upper index bound for a cursor. For use with (page 959) cursor.hint() (page 958) cursor.min() Species an inclusive lower index bound for a cursor. For use with (page 961) cursor.hint() (page 958) cursor.next() Returns the next document in a cursor. (page 962) cursor.objsLeftInBatch() Returns the number of documents left in the current cursor batch. (page 963) cursor.readPref() Species a read preference to a cursor to control how the client directs queries to a (page 963) replica set. cursor.showDiskLoc()Returns a cursor with modied documents that include the on- disk location of the (page 963) document. cursor.size() Returns a count of the documents in the cursor after applying skip() (page 964) (page 964) and limit() (page 959) methods. cursor.skip() Returns a cursor that begins returning results only after passing or skipping a (page 964) number of documents. cursor.snapshot() Forces the cursor to use the index on the _id eld. Ensures that the cursor returns (page 964) each document, with regards to the value of the _id eld, only once. cursor.sort() Returns results ordered according to a sort specication. (page 965) cursor.toArray() Returns an array that contains all documents returned by the cursor. (page 966)
cursor.addOption()
cursor.addOption(<ag>) Use the cursor.addOption() (page 951) method on a cursor to add OP_QUERY wire protocol ags, such as the tailable ag, to change the behavior of queries.
951
Parameters ag OP_QUERY wire protocol ag. See MongoDB wire protocol for more information on MongoDB Wire Protocols and the OP_QUERY ags. For the mongo (page 1036) shell, you can use the cursor ags listed in the Cursor Flags (page 177) section. For the driver-specic list, see your driver documentation (page 555). The following example in the mongo (page 1036) shell adds the DBQuery.Option.tailable ag and the DBQuery.Option.awaitData ag to ensure that the query returns a tailable cursor:
var t = db.myCappedCollection; var cursor = t.find().addOption(DBQuery.Option.tailable). addOption(DBQuery.Option.awaitData)
This sequence of operations creates a cursor that will wait for few seconds after returning the full result set so that it can capture and return additional data added during the query. Warning: Adding incorrect wire protocol ags can cause problems and/or extra server load.
cursor.batchSize()
cursor.batchSize() The batchSize() (page 952) method species the number of documents to return in each batch of the response from the MongoDB instance. In most cases, modifying the batch size will not affect the user or the application since the mongo (page 1036) shell and most drivers (page 555) return results as if MongoDB returned a single batch. The batchSize() (page 952) method takes the following parameter: Parameters size The number of documents to return per batch. Do not use a batch size of 1. Note: Specifying 1 or a negative number is analogous to using the limit() (page 959) method. Consider the following example of the batchSize() (page 952) method in the mongo (page 1036) shell:
db.inventory.find().batchSize(10)
This operation will set the batch size for the results of a query (i.e. find() (page 924)) to 10. The effects of this operation do not affect the output in the mongo (page 1036) shell, which always iterates over the rst 20 documents.
cursor.count()
cursor.count() The count() (page 952) method counts the number of documents referenced by a cursor. Append the count() (page 952) method to a find() (page 924) query to return the number of matching documents, as in the following prototype:
db.collection.find().count()
This operation does not actually perform the find() (page 924); instead, the operation counts the results that would be returned by the find() (page 924). The count() (page 952) can accept the following argument: 952 Chapter 63. MongoDB Interface
Parameters applySkipLimit (boolean) Optional. Species whether to consider the effects of the cursor.skip() (page 964) and cursor.limit() (page 959) methods in the count. By default, the count() (page 952) method ignores the effects of the cursor.skip() (page 964) and cursor.limit() (page 959). Set applySkipLimit to true to consider the effect of these methods. See also: cursor.size() (page 964) MongoDB also provides the shell wrapper db.collection.count() (page 919) for the db.collection.find().count() construct. Consider the following examples of the count() (page 952) method: Count the number of all documents in the orders collection:
db.orders.find().count()
db.orders.find( { ord_dt: { $gt: new Date(01/01/2012) } } ).count()
Count the number of the documents in the orders collection with the eld ord_dt greater than new Date(01/01/2012) taking into account the effect of the limit(5):
db.orders.find( { ord_dt: { $gt: new Date(01/01/2012) } } ).limit(5).count(true)
cursor.explain()
Denition cursor.explain() The cursor.explain() (page 953) method provides information on the query plan. The query plan is the plan the server uses to nd the matches for a query. This information may be useful when optimizing a query. Parameters verbose (boolean) Species the level of detail to include in the output. If true or 1, includes the allPlans and oldPlan elds in the output. Returns A document that describes the process used to return the query results. Retrieve the query plan by appending explain() (page 953) to a find() (page 924) query, as in the following example:
db.products.find().explain()
explain (page 953) runs the actual query to determine the result. Although there are some differences between running the query with explain (page 953) and running without, generally, the performance will be similar between the two. So, if the query is slow, the explain (page 953) operation is also slow. Additionally, the explain (page 953) operation reevaluates a set of candidate query plans, which may cause the explain (page 953) operation to perform differently than a normal query. As a result, these operations generally provide an accurate account of how MongoDB would perform the query, but do not reect the length of these queries. To determine the performance of a particular index, you can use hint() (page 958) and in conjunction with explain() (page 953), as in the following example:
953
db.products.find().hint( { type: 1 } ).explain()
When you run explain (page 953) with hint() (page 958), the query optimizer does not reevaluate the query plans. Note: In some situations, the explain() (page 953) operation may differ from the actual query plan used by MongoDB in a normal query. The explain() (page 953) operation evaluates the set of query plans and reports on the winning plan for the query. In normal operations the query optimizer caches winning query plans and uses them for similar related queries in the future. As a result MongoDB may sometimes select query plans from the cache that are different from the plan displayed using explain (page 953). See also: $explain (page 802) Optimization Strategies for MongoDB (page 555) page for information regarding optimization strategies. Analyze Performance of Database Operations (page 99) tutorial for information regarding the database prole. Current Operation Reporting (page 971) Output for Queries on Sharded Collections This section explains output for queries on sharded collections. For queries on non-sharded collections, see Core Explain Output (page 955). For queries on sharded collections, explain() (page 953) returns information for each shard the query accesses.
{ "cursor" : "<Cursor Type and Index>", "isMultiKey" : <boolean>, "n" : <num>, "nscannedObjects" : <num>, "nscanned" : <num>, "nscannedObjectsAllPlans" : <num>, "nscannedAllPlans" : <num>, "scanAndOrder" : <boolean>, "indexOnly" : <boolean>, "nYields" : <num>, "nChunkSkips" : <num>, "millis" : <num>, "indexBounds" : { <index bounds> }, "allPlans" : [ { "cursor" : "<Cursor Type and Index>", "n" : <num>, "nscannedObjects" : <num>, "nscanned" : <num>, "indexBounds" : { <index bounds> } }, ... ], "oldPlan" : { "cursor" : "<Cursor Type and Index>", "indexBounds" : { <index bounds> } } "server" : "<host:port>", }
954
$or Queries Queries with $or (page 770) operator execute each clause of the $or (page 770) expression in parallel and can use separate indexes on the individual clauses. If the query uses indexes on any or all of the querys clause, explain() (page 953) contains output (page 955) for each clause as well as the cumulative data for the entire query:
{ "clauses" : [ { <core explain output> }, { <core explain output> }, ... ], "n" : <num>, "nscannedObjects" : <num>, "nscanned" : <num>, "nscannedObjectsAllPlans" : <num>, "nscannedAllPlans" : <num>, "millis" : <num>, "server" : "<host:port>" }
Sharded Collections For queries on a sharded collection, the output contains the Core Explain Output (page 955) for each accessed shard and cumulative shard information (page 957):
{ "clusteredType" : "<Shard Access Type>", "shards" : { "<shard1>" : [ { <core explain output> } ], "<shard2>" : [ { <core explain output> } ], ... }, "millisShardTotal" : <num>, "millisShardAvg" : <num>, "numQueries" : <num>, "numShards" : <num>, "cursor" : "<Cursor Type and Index>", "n" : <num>, "nChunkSkips" : <num>, "nYields" : <num>, "nscanned" : <num>, "nscannedAllPlans" : <num>, "nscannedObjects" : <num>, "nscannedObjectsAllPlans" : <num>, "millis" : <num> }
955
Core Explain Output explain.cursor cursor (page 956) is a string that reports the type of cursor used by the query operation: BasicCursor indicates a full collection scan. BtreeCursor indicates that the query used an index. The cursor includes name of the index. When a query uses an index, the output of explain() (page 953) includes indexBounds (page 957) details. GeoSearchCursor indicates that the query used a geospatial index. explain.isMultiKey isMultiKey (page 956) is a boolean. When true, the query uses a multikey index (page 332), where one of the elds in the index holds an array. explain.n n (page 956) is a number that reects the number of documents that match the query selection criteria. explain.nscannedObjects Species the total number of documents scanned during the query. The nscannedObjects (page 956) may be lower than nscanned (page 956), such as if the index covers (page 342) a query. See indexOnly (page 956). Additionally, the nscannedObjects (page 956) may be lower than nscanned (page 956) in the case of multikey index on an array eld with duplicate documents. explain.nscanned Species the total number of documents or index entries scanned during the database operation. You want n (page 956) and nscanned (page 956) to be close in value as possible. The nscanned (page 956) value may be higher than the nscannedObjects (page 956) value, such as if the index covers (page 342) a query. See indexOnly (page 956). explain.nscannedObjectsAllPlans New in version 2.2. nscannedObjectsAllPlans (page 956) is a number that reects the total number of documents scanned for all query plans during the database operation. explain.nscannedAllPlans New in version 2.2. nscannedAllPlans (page 956) is a number that reects the total number of documents or index entries scanned for all query plans during the database operation. explain.scanAndOrder scanAndOrder (page 956) is a boolean that is true when the query cannot use the index for returning sorted results. When true, MongoDB must sort the documents after it retrieves them from either an index cursor or a cursor that scans the entire collection. explain.indexOnly indexOnly (page 956) is a boolean value that returns true when the query is covered (page 342) by the index indicated in the cursor (page 956) eld. When an index covers a query, MongoDB can both match the query conditions (page 168) and return the results using only the index because: all the elds in the query (page 168) are part of that index, and all the elds returned in the results set are in the same index. explain.nYields nYields (page 956) is a number that reects the number of times this query yielded the read lock to allow waiting writes execute.
956
explain.nChunkSkips nChunkSkips (page 956) is a number that reects the number of documents skipped because of active chunk migrations in a sharded system. Typically this will be zero. A number greater than zero is ok, but indicates a little bit of inefciency. explain.millis millis (page 957) is a number that reects the time in milliseconds to complete the query. explain.indexBounds indexBounds (page 957) is a document that contains the lower and upper index key bounds. This eld resembles one of the following:
"indexBounds" : { "start" : { <index key1> : <value>, ... "end" : { <index key1> : <value>, ... } }, "indexBounds" : { "<field>" : [ [ <lower bound>, <upper bound> ] ], ... } },
explain.allPlans allPlans (page 957) is an array that holds the list of plans the query optimizer runs in order to select the index for the query. Displays only when the <verbose> parameter to explain() (page 953) is true or 1. explain.oldPlan New in version 2.2. oldPlan (page 957) is a document value that contains the previous plan selected by the query optimizer for the query. Displays only when the <verbose> parameter to explain() (page 953) is true or 1. explain.server New in version 2.2. server (page 957) is a string that reports the MongoDB server. $or Query Output explain.clauses clauses (page 957) is an array that holds the Core Explain Output (page 955) information for each clause of the $or (page 770) expression. clauses (page 957) is only included when the clauses in the $or (page 770) expression use indexes. Sharded Collections Output explain.clusteredType clusteredType (page 957) is a string that reports the access pattern for shards. The value is: ParallelSort, if the mongos (page 1032) queries shards in parallel. SerialServer, if the mongos (page 1032) queries shards sequentially. explain.shards shards (page 957) contains elds for each shard in the cluster accessed during the query. Each eld holds the Core Explain Output (page 955) for that shard. explain.millisShardTotal millisShardTotal (page 957) is a number that reports the total time in milliseconds for the query to run on the shards.
957
explain.millisShardAvg millisShardAvg (page 957) is a number that reports the average time in millisecond for the query to run on each shard. explain.numQueries numQueries (page 958) is a number that reports the total number of queries executed. explain.numShards numShards (page 958) is a number that reports the total number of shards queried.
cursor.forEach()
cursor.forEach(<function>) The forEach() (page 958) method iterates the cursor to apply a JavaScript <function> to each document from the cursor. The forEach() (page 958) method accepts the following argument: Parameters <function> JavaScript function to apply to each document from the cursor. The <function> signature includes a single argument that is passed the current document to process. The following example invokes the forEach() (page 958) method on the cursor returned by find() (page 924) to print the name of each user in the collection:
db.users.find().forEach( function(myDoc) { print( "user: " + myDoc.name ); } );
See also: cursor.map() (page 959) for similar functionality.

cursor.hasNext()
cursor.hasNext() Returns Boolean. cursor.hasNext() (page 958) returns true if the cursor returned by the db.collection.find() (page 924) query can iterate further to return more documents.
cursor.hint()
cursor.hint(index) Arguments index The index to hint or force MongoDB to use when performing the query. Specify the index either by the index name or by the index specication document. See Index Specication Documents (page 192) for information on index specication documents. Call this method on a query to override MongoDBs default index selection and query optimization process. Use db.collection.getIndexes() (page 930) to return the list of current indexes on a collection. Consider the following operation:
958
db.users.find().hint( { age: 1 } )
This operation returns all documents in the collection named users using the index on the age eld. You can also specify the index using the index name:
db.users.find().hint( "age_1" )
See also: $hint (page 803)

cursor.limit()
cursor.limit() Use the cursor.limit() (page 959) method on a cursor to specify the maximum number of documents the cursor will return. cursor.limit() (page 959) is analogous to the LIMIT statement in a SQL database. Note: You must apply cursor.limit() (page 959) to the cursor before retrieving any documents from the database. Use cursor.limit() (page 959) to maximize performance and prevent MongoDB from returning more results than required for processing. A cursor.limit() (page 959) value of 0 (e.g. .limit(0) (page 959)) is equivalent to setting no limit.
cursor.map()
cursor.map(function) Parameters function function to apply to each document visited by the cursor. Apply function to each document visited by the cursor, and collect the return values from successive application into an array. Consider the following example:
db.users.find().map( function(u) { return u.name; } );
See also: cursor.forEach() (page 958) for similar functionality.

cursor.max()
cursor.max() The max() (page 959) method species the exclusive upper bound for a specic index in order to constrain the results of find() (page 924). max() (page 959) provides a way to specify an upper bound on compound key indexes. max() (page 959) takes the following parameter: Parameters indexBounds (document) Species the exclusive upper bound for the index keys. The indexBounds parameter has the following prototype form:
959
{ field1: <max value>, field2: <max value2> ... fieldN:<max valueN>}
The elds correspond to all the keys of a particular index in order. You can explicitly specify the particular index with the hint() (page 958) method. Otherwise, mongod (page 1021) selects the index using the elds in the indexBounds; however, if multiple indexes exist on same elds with different sort orders, the selection of the index may be ambiguous. See also: min() (page 961). Consider the following example of max() (page 959), which assumes a collection named products that holds the following documents:
{ { { { { { { { { { "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" : : : : : : : : : : 6, "item" : "apple", "type" : "cortland", "price" : 1.29 } 2, "item" : "apple", "type" : "fuji", "price" : 1.99 } 1, "item" : "apple", "type" : "honey crisp", "price" : 1.99 } 3, "item" : "apple", "type" : "jonagold", "price" : 1.29 } 4, "item" : "apple", "type" : "jonathan", "price" : 1.29 } 5, "item" : "apple", "type" : "mcintosh", "price" : 1.29 } 7, "item" : "orange", "type" : "cara cara", "price" : 2.99 } 10, "item" : "orange", "type" : "navel", "price" : 1.39 } 9, "item" : "orange", "type" : "satsuma", "price" : 1.99 } 8, "item" : "orange", "type" : "valencia", "price" : 0.99 }

{ { { { "_id" : 1 } "item" : 1, "type" : 1 } "item" : 1, "type" : -1 } "price" : 1 }
Using the ordering of { item: 1, type: 1 } index, max() (page 959) limits the query to the documents that are below the bound of item equal to apple and type equal to jonagold:
db.products.find().max( { item: apple, type: jonagold } ).hint( { item: 1, type: 1 } )
The query returns the following documents:

{ "_id" : 6, "item" : "apple", "type" : "cortland", "price" : 1.29 } { "_id" : 2, "item" : "apple", "type" : "fuji", "price" : 1.99 } { "_id" : 1, "item" : "apple", "type" : "honey crisp", "price" : 1.99 }
If the query did not explicitly specify the index with the hint() (page 958) method, it is ambiguous as to whether mongod (page 1021) would select the { item: 1, type: 1 } index ordering or the { item: 1, type: -1 } index ordering. Using the ordering of the index { price: 1 }, max() (page 959) limits the query to the documents that are below the index key bound of price equal to 1.99 and min() (page 961) limits the query to the documents that are at or above the index key bound of price equal to 1.39:
db.products.find().min( { price: 1.39 } ).max( { price: 1.99 } ).hint( { price: 1 } )

{ { { { { "_id" "_id" "_id" "_id" "_id" : : : : : 6, "item" : "apple", "type" : 4, "item" : "apple", "type" : 5, "item" : "apple", "type" : 3, "item" : "apple", "type" : 10, "item" : "orange", "type" "cortland", "price" : 1.29 } "jonathan", "price" : 1.29 } "mcintosh", "price" : 1.29 } "jonagold", "price" : 1.29 } : "navel", "price" : 1.39 }
960
Note: Because max() (page 959) requires an index on a eld, and forces the query to use this index, you may prefer the $lt (page 766) operator for the query if possible. Consider the following example:
db.products.find( { _id: 7 } ).max( { price: 1.39 } )
The query will use the index on the price eld, even if the index on _id may be better. max() (page 959) exists primarily to support the mongos (page 1032) (sharding) process. If you use max() (page 959) with min() (page 961) to specify a range, the index bounds specied in min() (page 961) and max() (page 959) must both refer to the keys of the same index. max() (page 959) is a shell wrapper around the special operator $max (page 804).
cursor.min()
cursor.min() The min() (page 961) method species the inclusive lower bound for a specic index in order to constrain the results of find() (page 924). min() (page 961) provides a way to specify lower bounds on compound key indexes. min() (page 961) takes the following parameter: Parameters indexBounds (document) Species the inclusive lower bound for the index keys. The indexBounds parameter has the following prototype form:
{ field1: <min value>, field2: <min value2>, fieldN:<min valueN> }
The elds correspond to all the keys of a particular index in order. You can explicitly specify the particular index with the hint() (page 958) method. Otherwise, MongoDB selects the index using the elds in the indexBounds; however, if multiple indexes exist on same elds with different sort orders, the selection of the index may be ambiguous. See also: max() (page 959). Consider the following example of min() (page 961), which assumes a collection named products that holds the following documents:
{ { { { { { { { { { "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" "_id" : : : : : : : : : : 6, "item" : "apple", "type" : "cortland", "price" : 1.29 } 2, "item" : "apple", "type" : "fuji", "price" : 1.99 } 1, "item" : "apple", "type" : "honey crisp", "price" : 1.99 } 3, "item" : "apple", "type" : "jonagold", "price" : 1.29 } 4, "item" : "apple", "type" : "jonathan", "price" : 1.29 } 5, "item" : "apple", "type" : "mcintosh", "price" : 1.29 } 7, "item" : "orange", "type" : "cara cara", "price" : 2.99 } 10, "item" : "orange", "type" : "navel", "price" : 1.39 } 9, "item" : "orange", "type" : "satsuma", "price" : 1.99 } 8, "item" : "orange", "type" : "valencia", "price" : 0.99 }
961
{ { { {
"_id" : 1 } "item" : 1, "type" : 1 } "item" : 1, "type" : -1 } "price" : 1 }
Using the ordering of { item: 1, type: 1 } index, min() (page 961) limits the query to the documents that are at or above the index key bound of item equal to apple and type equal to jonagold, as in the following:
db.products.find().min( { item: apple, type: jonagold } ).hint( { item: 1, type: 1 } )

{ { { { { { { "_id" "_id" "_id" "_id" "_id" "_id" "_id" : : : : : : : 3, "item" : "apple", "type" : "jonagold", "price" : 1.29 } 4, "item" : "apple", "type" : "jonathan", "price" : 1.29 } 5, "item" : "apple", "type" : "mcintosh", "price" : 1.29 } 7, "item" : "orange", "type" : "cara cara", "price" : 2.99 } 10, "item" : "orange", "type" : "navel", "price" : 1.39 } 9, "item" : "orange", "type" : "satsuma", "price" : 1.99 } 8, "item" : "orange", "type" : "valencia", "price" : 0.99 }
If the query did not explicitly specify the index with the hint() (page 958) method, it is ambiguous as to whether mongod (page 1021) would select the { item: 1, type: 1 } index ordering or the { item: 1, type: -1 } index ordering. Using the ordering of the index { price: 1 }, min() (page 961) limits the query to the documents that are at or above the index key bound of price equal to 1.39 and max() (page 959) limits the query to the documents that are below the index key bound of price equal to 1.99:
db.products.find().min( { price: 1.39 } ).max( { price: 1.99 } ).hint( { price: 1 } )

{ { { { { "_id" "_id" "_id" "_id" "_id" : : : : : 6, "item" : "apple", "type" : 4, "item" : "apple", "type" : 5, "item" : "apple", "type" : 3, "item" : "apple", "type" : 10, "item" : "orange", "type" "cortland", "price" : 1.29 } "jonathan", "price" : 1.29 } "mcintosh", "price" : 1.29 } "jonagold", "price" : 1.29 } : "navel", "price" : 1.39 }
Note: Because min() (page 961) requires an index on a eld, and forces the query to use this index, you may prefer the $gte (page 765) operator for the query if possible. Consider the following example:
db.products.find( { _id: 7 } ).min( { price: 1.39 } )
The query will use the index on the price eld, even if the index on _id may be better. min() (page 961) exists primarily to support the mongos (page 1032) (sharding) process. If you use min() (page 961) with max() (page 959) to specify a range, the index bounds specied in min() (page 961) and max() (page 959) must both refer to the keys of the same index. min() (page 961) is a shell wrapper around the special operator $min (page 804).
cursor.next()
cursor.next() 962 Chapter 63. MongoDB Interface
Returns The next document in the cursor returned by the db.collection.find() (page 924) method. See cursor.hasNext() (page 958) related functionality.
cursor.objsLeftInBatch()
cursor.objsLeftInBatch() cursor.objsLeftInBatch() (page 963) returns the number of documents remaining in the current batch. The MongoDB instance returns response in batches. To retrieve all the documents from a cursor may require multiple batch responses from the MongoDB instance. When there are no more documents remaining in the current batch, the cursor will retrieve another batch to get more documents until the cursor exhausts.
cursor.readPref()
cursor.readPref() Parameters mode (string) Read preference mode. tagSet (array) Optional. Array of tag set objects. Append the readPref() (page 963) to a cursor to control how the client will route the query will route to members of the replica set. The mode string should be one of: primary (page 403) primaryPreferred (page 403) secondary (page 403) secondaryPreferred (page 404) nearest (page 404) See the tag sets (page 404) documentation for more information on using tag sets to provide custom read preference modes. Note: You must apply cursor.readPref() (page 963) to the cursor before retrieving any documents from the database.
cursor.showDiskLoc()
cursor.showDiskLoc() Returns A modied cursor object that contains documents with appended information that describes the on-disk location of the document. See also: $showDiskLoc (page 806) for related functionality.
963
cursor.size()
cursor.size() Returns A count of the number of documents that match the db.collection.find() (page 924) query after applying any cursor.skip() (page 964) and cursor.limit() (page 959) methods.
cursor.skip()
cursor.skip() Call the cursor.skip() (page 964) method on a cursor to control where MongoDB begins returning results. This approach may be useful in implementing paged results. Note: You must apply cursor.skip() (page 964) to the cursor before retrieving any documents from the database. Consider the following JavaScript function as an example of the sort function:
function printStudents(pageNumber, nPerPage) { print("Page: " + pageNumber); db.students.find().skip((pageNumber-1)*nPerPage).limit(nPerPage).forEach( function(student) { }
The cursor.skip() (page 964) method is often expensive because it requires the server to walk from the beginning of the collection or index to get the offset or skip position before beginning to return result. As offset (e.g. pageNumber above) increases, cursor.skip() (page 964) will become slower and more CPU intensive. With larger collections, cursor.skip() (page 964) may become IO bound. Consider using range-based pagination for these kinds of tasks. That is, query for a range of objects, using logic within the application to determine the pagination rather than the database itself. This approach features better index utilization, if you do not need to easily jump to a specic page.
cursor.snapshot()
cursor.snapshot() Append the cursor.snapshot() (page 964) method to a cursor to toggle the snapshot mode. This ensures that the query will not return a document multiple times, even if intervening write operations result in a move of the document due to the growth in document size. Warning: You must apply cursor.snapshot() (page 964) to the cursor before retrieving any documents from the database. You can only use snapshot() (page 964) with unsharded collections. The snapshot() (page 964) does not guarantee isolation from insertion or deletions. The cursor.snapshot() (page 964) traverses the index on the _id eld. As such, snapshot() (page 964) cannot be used with sort() (page 965) or hint() (page 958). Queries with results of less than 1 megabyte are effectively implicitly snapshotted.
964
cursor.sort()
cursor.sort(sort) Parameters sort A document whose elds specify the attributes on which to sort the result set. Append the sort() (page 965) method to a cursor to control the order that the query returns matching documents. For each eld in the sort document, if the elds corresponding value is positive, then sort() (page 965) returns query results in ascending order for that attribute: if the elds corresponding value is negative, then sort() (page 965) returns query results in descending order. Note: You must apply cursor.limit() (page 959) to the cursor before retrieving any documents from the database. Consider the following example:
db.collection.find().sort( { age: -1 } );
Here, the query returns all documents in collection sorted by the age eld in descending order. Specify a value of negative one (e.g. -1), as above, to sort in descending order or a positive value (e.g. 1) to sort in ascending order. Unless you have an index for the specied key pattern, use cursor.sort() (page 965) in conjunction with cursor.limit() (page 959) to avoid requiring MongoDB to perform a large, in-memory sort. cursor.limit() (page 959) increases the speed and reduces the amount of memory required to return this query by way of an optimized algorithm. Warning: The sort function requires that the entire sort be able to complete within 32 megabytes. When the sort option consumes more than 32 megabytes, MongoDB will return an error. Use cursor.limit() (page 959), or create an index on the eld that youre sorting to avoid this error. The $natural (page 807) parameter returns items according to their order on disk. Consider the following query:
db.collection.find().sort( { $natural: -1 } )
This will return documents in the reverse of the order on disk. Typically, the order of documents on disks reects insertion order, except when documents move internal because of document growth due to update operations. When comparing values of different BSON types, MongoDB uses the following comparison order, from lowest to highest: 1.MinKey (internal type) 2.Null 3.Numbers (ints, longs, doubles) 4.Symbol, String 5.Object 6.Array 7.BinData 8.ObjectID 9.Boolean
965
10.Date, Timestamp 11.Regular Expression 12.MaxKey (internal type) Note: MongoDB treats some types as equivalent for comparison purposes. For instance, numeric types undergo conversion before comparison.
cursor.toArray()
cursor.toArray() The toArray() (page 966) method returns an array that contains all the documents from a cursor. The method iterates completely the cursor, loading all the documents into RAM and exhausting the cursor. Returns An array of documents. Consider the following example that applies toArray() (page 966) to the cursor returned from the find() (page 924) method:
var allProductsArray = db.products.find().toArray(); if (allProductsArray.length > 0) { printjson (allProductsArray[0]); }
The variable allProductsArray holds the array of documents returned by toArray() (page 966).
63.3.3 Database
Database Methods Name db.addUser() (page 967) db.auth() (page 968) db.cloneCollection() (page 969) db.cloneDatabase() (page 969) db.commandHelp() (page 969) db.copyDatabase() (page 970) db.createCollection() (page 970) db.currentOp() (page 971) db.dropDatabase() (page 975) db.eval() (page 975) db.fsyncLock() (page 977) db.fsyncUnlock() (page 977) db.getCollection() (page 978) db.getCollectionNames() (page 978) db.getLastError() (page 978) db.getLastErrorObj() (page 978) db.getMongo() (page 978) db.getName() (page 978) db.getPrevError() (page 979) db.getProfilingLevel() (page 979) db.getProfilingStatus() (page 979)
Description Adds a user to a database, and allows administrators to congure the users Authenticates a user to a database. Copies data directly between MongoDB instances. Wraps cloneCollec Copies a database from a remote host to the current host. Wraps clone (p Returns help information for a database command. Copies a database to another database on the current host. Wraps copydb Creates a new collection. Commonly used to create a capped collection. Reports the current in-progress operations. Removes the current database. Passes a JavaScript function to the mongod (page 1021) instance for serve Flushes writes to disk and locks the database to prevent write operations an Allows writes to continue on a database locked with db.fsyncLock() ( Returns a collection object. Used to access collections with names that are Lists all collections in the current database. Checks and returns the status of the last operation. Wraps getLastErro Returns the status document for the last operation. Wraps getLastErro Returns the Mongo() (page 1007) connection object for the current conne Returns the name of the current database. Returns a status document containing all errors since the last error reset. W Returns the current proling level for database operations. Returns a document that reects the current proling level and the proling
966
Name db.getReplicationInfo() (page 979) db.getSiblingDB() (page 980) db.help() (page 980) db.hostInfo() (page 980) db.isMaster() (page 981) db.killOp() (page 981) db.listCommands() (page 981) db.loadServerScripts() (page 982) db.logout() (page 982) db.printCollectionStats() (page 982) db.printReplicationInfo() (page 983) db.printShardingStatus() (page 983) db.printSlaveReplicationInfo() (page 983) db.removeUser() (page 983) db.repairDatabase() (page 984) db.resetError() (page 984) db.runCommand() (page 984) db.serverBuildInfo() (page 984) db.serverStatus() (page 984) db.setProfilingLevel() (page 985) db.shutdownServer() (page 985) db.stats() (page 986) db.version() (page 986)
db.addUser()
Table 63.2 continued from previous page Description Returns a document with replication statistics. Provides access to the specied database. Displays descriptions of common db object methods. Returns a document with information about the system MongoDB runs on. Returns a document that reports the state of the replica set. Terminates a specied operation. Displays a list of common database commands. Loads all scripts in the system.js collection for the current database int Ends an authenticated session. Prints statistics from every collection. Wraps db.collection.stats( Prints a report of the status of the replica set from the perspective of the pri Prints a report of the sharding conguration and the chunk ranges. Prints a report of the status of the replica set from the perspective of the sec Removes a user from a database. Runs a repair routine on the current database. Resets the error message returned by db.getPrevError() (page 979) Runs a database command (page 807). Returns a document that displays the compilation parameters for the mong Returns a document that provides an overview of the state of the database p Modies the current level of database proling. Shuts down the current mongod (page 1021) or mongos (page 1032) proc Returns a document that reports on the state of the current database. Returns the version of the mongod (page 1021) instance.
db.addUser() Use db.addUser() (page 967) to add privilege documents to the system.users (page 160) collection in a database, which creates database credentials in MongoDB. Changed in version 2.4: The schema of system.users (page 160) changed in 2.4 to accommodate a more sophisticated privilege model (page 155). In 2.4 db.addUser() (page 967) supports both forms of privilege documents. In MongoDB 2.4 you must pass db.addUser() (page 967) a document that contains a well-formed system.users (page 160) document. In MongoDB 2.2 pass arguments to db.addUser() (page 967) that describe user credentials (page 968). See system.users Privilege Documents (page 159) for more information about the form of the 2.4 privilege documents. Consider the following prototypes form for a db.addUser() (page 967) operations:
db.addUser( { user: "<user>", pwd: "<password>", roles: [<roles>] } )
This operation creates a system.users (page 160) document with a password using the pwd (page 160) eld In the following prototype, rather than specify a password directly, you can delegated the credential to another database using the userSource (page 161) eld:
db.addUser( { user: "<user>", userSource: "<database>", roles: [<roles>] } )
Example
967
To create and add a 2.4-style privilege document to system.users (page 160) to grant readWrite (page 156) privileges to a user named author with privileges, use the following operation:
db.addUser( { user: "author", pwd: "pass", roles: [ "readWrite" ] } )
If you want to store user credentials in a single users database, you can use delegated credentials (page 162), as in the following example:
db.addUser( { user: "author", userSource: "users", roles: [ "readWrite" ] } )
See also: Add a User to a Database (page 148) User Privilege Roles in MongoDB (page 155) system.users Privilege Documents (page 159) To create legacy (2.2. and earlier) privilege documents, db.addUser() (page 967) provides accepts the following parameters: Parameters user (string) Species the username. password (string) Species the corresponding password. readOnly (boolean) Optional. Defaults to false. Grants users a restricted privilege set that only allows the user to read the this database. Consider the following syntax:
db.addUser( "<username>", "<password>", { readOnly: <boolean> } )
Example To create and add a legacy (2.2. and earlier) privilege document with a user named guest and the password pass that has only readOnly privileges, use the following operation:
db.addUser( "guest", "pass", { readOnly: true } )
Note: The mongo (page 1036) shell excludes all db.addUser() (page 967) operations from the saved history. Deprecated since version 2.4: The roles parameter replaces the readOnly parameter for db.addUser() (page 967). 2.4 also adds the otherDBRoles (page 161) and userSource (page 161) elds to documents in the system.users (page 160) collection.
db.auth()
db.auth(username, password) Parameters username (string) Species an existing username with access privileges for this database. password (string) Species the corresponding password.
968
Allows a user to authenticate to the database from within the shell. Alternatively use mongo --username (page 1037) and --password (page 1037) to specify authentication credentials. Note: The mongo (page 1036) shell excludes all db.auth() (page 968) operations from the saved history.
db.cloneCollection()
db.cloneCollection(from, collection, query) db.cloneCollection() (page 969) provides a wrapper around cloneCollection (page 856) for copying data directly between MongoDB instances. db.cloneCollection() (page 969) does not allow you to clone a collection through a mongos (page 1032): you must connect directly to the mongod (page 1021) instance. Parameters from (string) A host name, of the MongoDB instance that holds the collection you wish to copy collection (string) A collection in the MongoDB instance that you want to copy. db.cloneCollection() (page 969) will only copy the collection with this name from database of the same name as the current database the remote MongoDB instance. If you want to copy a collection from a different database name you must use the cloneCollection (page 856) directly. query (document) Optional. A standard MongoDB query document (page 190) to limit the documents copied as part of the db.cloneCollection() (page 969) operation.
db.cloneDatabase()
db.cloneDatabase(hostname) Parameters hostname (string) Species the hostname to copy the current instance. Use this function to copy a database from a remote to the current database. The command assumes that the remote database has the same name as the current database. For example, to clone a database named importdb on a host named hostname, do
use importdb db.cloneDatabase("hostname");
New databases are implicitly created, so the current host does not need to have a database named importdb for this command to succeed. This function provides a wrapper around the MongoDB database command clone (page 857). The copydb (page 862) database command provides related functionality.
db.commandHelp()
db.commandHelp(command) Parameters command Species a database command name (page 807).
969
Returns Help text for the specied database command. See the database command reference (page 807) for full documentation of these commands.
db.copyDatabase()
db.copyDatabase(origin, destination, hostname) Parameters origin (database) Species the name of the database on the origin system. destination (database) Species the name of the database that you wish to copy the origin database into. hostname (origin) Indicate the hostname of the origin database host. Omit the hostname to copy from one name to another on the same server. Use this function to copy a specic database, named origin running on the system accessible via hostname into the local database named destination. The command creates destination databases implicitly when they do not exist. If you omit the hostname, MongoDB will copy data from one database into another on the same instance. This function provides a wrapper around the MongoDB database command copydb (page 862). The clone (page 857) database command provides related functionality. Example Given a database named records that you want to copy into a database named archive_records, you could use the following invocation of db.copyDatabase() (page 970):
db.copyDatabase(records, archive_records)
db.createCollection()
db.createCollection(name[, {capped: <boolean>, size: <value>, max <bytes>} ]) Parameters name (string) Species the name of a collection to create. capped (boolean) Optional. If this document is present, this command creates a capped collection. The capped argument is a document that contains the following three elds: capped Enables a collection cap. False by default. If enabled, you must specify a size parameter. size (bytes) If capped is true, size species a maximum size in bytes for the capped collection. When capped is false, you may use size to preallocate space. max (int) Optional. Species a maximum cap, in number of documents for capped collections. You must also specify size when specifying max. Options autoIndexId If capped is true you can specify false to disable the automatic index created on the _id eld. Before 2.2, the default value for autoIndexId was false. See _id Fields and Indexes on Capped Collections (page 1162) for more information.
970
Explicitly creates a new collection. Because MongoDB creates collections implicitly when referenced, this command is primarily used for creating new capped collections. In some circumstances, you may use this command to pre-allocate space for an ordinary collection. Capped collections have maximum size or document counts that prevent them from growing beyond maximum thresholds. All capped collections must specify a maximum size, but may also specify a maximum document count. MongoDB will remove older documents if a collection reaches the maximum size limit before it reaches the maximum document count. Consider the following example:
db.createCollection("log", { capped : true, size : 5242880, max : 5000 } )
This command creates a collection named log with a maximum size of 5 megabytes and a maximum of 5000 documents. The following command simply pre-allocates a 2 gigabyte, uncapped collection named people:
db.createCollection("people", { size: 2147483648 })
This command provides a wrapper around the database command create (page 863). See Capped Collections (page 558) for more information about capped collections.
db.currentOp()
Denition db.currentOp() Returns A document that reports in-progress operations for the database instance. The db.currentOp() (page 971) method can take no arguments or take the true argument, which returns a more verbose output, including idle connections and system operations. The following example uses the true argument:
db.currentOp("true")
db.currentOp() (page 971) is available only for users with administrative privileges. You can use db.killOp() (page 981) in conjunction with the opid (page 973) eld to terminate a currently running operation. The following JavaScript operations for the mongo (page 1036) shell lter the output of specic types of operations: Return all pending write operations:
db.currentOp().inprog.forEach( function(d){ if(d.waitingForLock && d.lockType != "read") printjson(d) })
Return the active write operation:

db.currentOp().inprog.forEach( function(d){ if(d.active && d.lockType == "write") printjson(d) })
Return all active read operations:
971
db.currentOp().inprog.forEach( function(d){ if(d.active && d.lockType == "read") printjson(d) })
Warning: Terminate running operations with extreme caution. Only use db.killOp() (page 981) to terminate operations initiated by clients and do not terminate internal database operations.
Example The following is an example of db.currentOp() (page 971) output. If you specify the true argument, db.currentOp() (page 971) returns more verbose output.
{ "inprog": [ { "opid" : 3434473, "active" : <boolean>, "secs_running" : 0, "op" : "<operation>", "ns" : "<database>.<collection>", "query" : { }, "client" : "<host>:<outgoing>", "desc" : "conn57683", "threadId" : "0x7f04a637b700", "connectionId" : 57683, "locks" : { "^" : "w", "^local" : "W", "^<database>" : "W" }, "waitingForLock" : false, "msg": "<string>" "numYields" : 0, "progress" : { "done" : <number>, "total" : <number> } "lockStats" : { "timeLockedMicros" : { "R" : NumberLong(), "W" : NumberLong(), "r" : NumberLong(), "w" : NumberLong() }, "timeAcquiringMicros" : { "R" : NumberLong(), "W" : NumberLong(), "r" : NumberLong(), "w" : NumberLong() } } }, ] }
972
Output Changed in version 2.2. The db.currentOp() (page 971) returns a document with an array named inprog. The inprog array contains a document for each in-progress operation. The elds that appear for a given operation depend on the kind of operation and its state. currentOp.opid Holds an identier for the operation. You can pass this value to db.killOp() (page 981) in the mongo (page 1036) shell to terminate the operation. currentOp.active A boolean value, that is true if the operation has started or false if the operation is queued and waiting for a lock to run. active (page 973) may be true even if the operation has yielded to another operation. currentOp.secs_running The duration of the operation in seconds. MongoDB calculates this value by subtracting the current time from the start time of the operation. If the operation is not running, (i.e. if active (page 973) is false,) this eld may not appear in the output of db.currentOp() (page 971). currentOp.op A string that identies the type of operation. The possible values are: insert query update remove getmore command currentOp.ns The namespace the operation targets. MongoDB forms namespaces using the name of the database and the name of the collection. currentOp.query A document containing the current operations query. The document is empty for operations that do not have queries: getmore, insert, and command. currentOp.client The IP address (or hostname) and the ephemeral port of the client connection where the operation originates. If your inprog array has operations from many different clients, use this string to relate operations to clients. For some commands, including findAndModify (page 826) and db.eval() (page 975), the client will be 0.0.0.0:0, rather than an actual client. currentOp.desc A description of the client. This string includes the connectionId (page 973). currentOp.threadId An identier for the thread that services the operation and its connection. currentOp.connectionId An identier for the connection where the operation originated. currentOp.locks New in version 2.2.
973
The locks (page 973) document reports on the kinds of locks the operation currently holds. The following kinds of locks are possible: currentOp.locks.^ ^ (page 974) reports on the use of the global lock :for the program:mongod instance. All operations must hold the :global lock for some phases of operation. currentOp.locks.^local ^local (page 974) reports on the lock for the local database. MongoDB uses the local database for a number of operations, but the most frequent use of the local database is for the oplog used in replication. currentOp.locks.^<database> locks.^<database> (page 974) reports on the lock state for the database that this operation targets. locks (page 973) replaces lockType in earlier versions. currentOp.lockType Changed in version 2.2: The locks (page 973) replaced the lockType (page 974) eld in 2.2. Identies the type of lock the operation currently holds. The possible values are: read write currentOp.waitingForLock Returns a boolean value. waitingForLock (page 974) is true if the operation is waiting for a lock and false if the operation has the required lock. currentOp.msg The msg (page 974) provides a message that describes the status and progress of the operation. In the case of indexing or mapReduce operations, the eld reports the completion percentage. currentOp.progress Reports on the progress of mapReduce or indexing operations. The progress (page 974) elds corresponds to the completion percentage in the msg (page 974) eld. The progress (page 974) species the following information: currentOp.progress.done Reports the number completed. currentOp.progress.total Reports the total number. currentOp.killed Returns true if mongod (page 1021) instance is in the process of killing the operation. currentOp.numYields numYields (page 974) is a counter that reports the number of times the operation has yielded to allow other operations to complete. Typically, operations yield when they need access to data that MongoDB has not yet fully read into memory. This allows other operations that have data in memory to complete quickly while MongoDB reads in data for the yielding operation. currentOp.lockStats New in version 2.2. The lockStats (page 974) document reects the amount of time the operation has spent both acquiring and holding locks. lockStats (page 974) reports data on a per-lock type, with the following possible lock types: R represents the global read lock,
974
W represents the global write lock, r represents the database specic read lock, and w represents the database specic write lock. currentOp.timeLockedMicros The timeLockedMicros (page 975) document reports the amount of time the operation has spent holding a specic lock. For operations that require more than one lock, like those that lock the local database to update the oplog, then the values in this document can be longer than this value may be longer than the total length of the operation (i.e. secs_running (page 973).) currentOp.timeLockedMicros.R Reports the amount of time in microseconds the operation has held the global read lock. currentOp.timeLockedMicros.W Reports the amount of time in microseconds the operation has held the global write lock. currentOp.timeLockedMicros.r Reports the amount of time in microseconds the operation has held the database specic read lock. currentOp.timeLockedMicros.w Reports the amount of time in microseconds the operation has held the database specic write lock. currentOp.timeAcquiringMicros The timeAcquiringMicros (page 975) document reports the amount of time the operation has spent waiting to acquire a specic lock. currentOp.timeAcquiringMicros.R Reports the mount of time in microseconds the operation has waited for the global read lock. currentOp.timeAcquiringMicros.W Reports the mount of time in microseconds the operation has waited for the global write lock. currentOp.timeAcquiringMicros.r Reports the mount of time in microseconds the operation has waited for the database specic read lock. currentOp.timeAcquiringMicros.w Reports the mount of time in microseconds the operation has waited for the database specic write lock.
db.dropDatabase()
db.dropDatabase() Removes the current database. Does not change the current database, so the insertion of any documents in this database will allocate a fresh set of data les.
db.eval()
db.eval(function, arguments) The db.eval() (page 975) provides the ability to run JavaScript code on the MongoDB server.
975
The helper db.eval() (page 975) in the mongo (page 1036) shell wraps the eval (page 823) command. Therefore, the helper method shares the characteristics and behavior of the underlying command with one exception: db.eval() (page 975) method does not support the nolock option. The method accepts the following parameters: Parameters function (JavaScript) A JavaScript function. The function need not take any arguments, as in the rst example, or may take arguments as in the second:
function () { // ... } function (arg1, arg2) { // ... }
arguments A list of corresponding arguments to pass to the specied JavaScript function if the function accepts arguments. Omit if the function does not take arguments. Consider the following example of the db.eval() (page 975) method:
db.eval( function(name, incAmount) { var doc = db.myCollection.findOne( { name : name } ); doc = doc || { name : name , num : 0 , total : 0 , avg : 0 }; doc.num++; doc.total += incAmount; doc.avg = doc.total / doc.num; db.myCollection.save( doc ); return doc; }, "eliot", 5 );
The db in the function refers to the current database. "eliot" is the argument passed to the function, and corresponds to the name argument. 5 is an argument to the function and corresponds to the incAmount eld. If you want to use the servers interpreter, you must run db.eval() (page 975). Otherwise, the mongo (page 1036) shells JavaScript interpreter evaluates functions entered directly into the shell. If an error occurs, db.eval() (page 975) throws an exception. Consider the following invalid function that uses the variable x without declaring it as an argument:
db.eval( function() { return x + x; }, 3 );
The statement will result in the following exception:

{ "errmsg" : "exception: JavaScript execution failed: ReferenceError: x is not defined near { "code" : 16722,
976
"ok" : 0 }
Warning: By default, db.eval() (page 975) takes a global write lock before evaluating the JavaScript function. As a result, db.eval() (page 975) blocks all other read and write operations to the database while the db.eval() (page 975) operation runs. Set nolock to true on the eval (page 823) command to prevent the eval (page 823) command from taking the global write lock before evaluating the JavaScript. nolock does not impact whether operations within the JavaScript code itself takes a write lock. Do not use db.eval() (page 975) for long running operations as db.eval() (page 975) blocks all other operations. Consider using other server side code execution options (page 560). You can not use db.eval() (page 975) with sharded data. In general, you should avoid using db.eval() (page 975) in sharded cluster; nevertheless, it is possible to use db.eval() (page 975) with non-sharded collections and databases stored in a sharded cluster. With authentication (page 1081) enabled, db.eval() (page 975) will fail during the operation if you do not have the permission to perform a specied task. Changed in version 2.4: You must have full admin access to run. Changed in version 2.4: The V8 JavaScript engine, which became the default in 2.4, allows multiple JavaScript operations to execute at the same time. Prior to 2.4, db.eval() (page 975) executed in a single thread. See also: Server-side JavaScript (page 560)
db.fsyncLock()
db.fsyncLock() Forces the mongod (page 1021) to ush pending all write operations to the disk and locks the entire mongod (page 1021) instance to prevent additional writes until the user releases the lock with the db.fsyncUnlock() (page 977) command. db.fsyncLock() (page 977) is an administrative command. This command provides a simple wrapper around a fsync (page 865) database command with the following syntax:
{ fsync: 1, lock: true }
This function locks the database and create a window for backup operations (page 41). Note: The database cannot be locked with db.fsyncLock() (page 977) while proling is enabled. You must disable proling before locking the database with db.fsyncLock() (page 977). Disable proling using db.setProfilingLevel() (page 985) as follows in the mongo (page 1036) shell:
db.fsyncUnlock()
db.fsyncUnlock() Unlocks a mongod (page 1021) instance to allow writes and reverses the operation of a db.fsyncLock() (page 977) operation. Typically you will use db.fsyncUnlock() (page 977) following a database backup operation (page 41). db.fsyncUnlock() (page 977) is an administrative command. 63.3. mongo Shell Methods 977
db.getCollection()
db.getCollection(name) Parameters name The name of a collection. Returns A collection. Use this command to obtain a handle on a collection whose name might interact with the shell itself, including collections with names that begin with _ or mirror the database commands (page 807).
db.getCollectionNames()
db.getCollectionNames() Returns An array containing all collections in the existing database.

db.getLastError()
db.getLastError() Returns The last error message string. Sets the level of write concern for conrming the success of write operations. See also: getLastError (page 831) for all options, Write Concern (page 398) for a conceptual overview, Write Operations (page 179) for information about all write operations in MongoDB, and Replica Set Write Concern (page 398) for special considerations related to write concern for replica sets.
db.getLastErrorObj()
db.getLastErrorObj() Returns A full document with status information.

db.getMongo()
db.getMongo() Returns The current database connection. db.getMongo() (page 978) runs when the shell initiates. Use this command to test that the mongo (page 1036) shell has a connection to the proper database instance.
db.getName()
db.getName() Returns the current database name.
978
db.getPrevError()
db.getPrevError() Returns A status document, containing the errors. Deprecated since version 1.6. This output reports all errors since the last time the database received a resetError (page 832) (also db.resetError() (page 984)) command. This method provides a wrapper around the getPrevError (page 832) command.
db.getProlingLevel()
db.getProfilingLevel() This method provides a wrapper around the database command profile (page 888) and returns the current proling level. Deprecated since version 1.8.4: Use db.getProfilingStatus() (page 979) for related functionality.
db.getProlingStatus()
db.getProfilingStatus() Returns The current profile (page 888) level and slowms (page 1085) setting.
db.getReplicationInfo()
Denition db.getReplicationInfo() Returns A document with the status of the replica status, using data polled from the oplog. Use this output when diagnosing issues with replication. Output db.getReplicationInfo.logSizeMB Returns the total size of the oplog in megabytes. This refers to the total amount of space allocated to the oplog rather than the current size of operations stored in the oplog. db.getReplicationInfo.usedMB Returns the total amount of space used by the oplog in megabytes. This refers to the total amount of space currently used by operations stored in the oplog rather than the total amount of space allocated. db.getReplicationInfo.errmsg Returns an error message if there are no entries in the oplog. db.getReplicationInfo.oplogMainRowCount Only present when there are no entries in the oplog. Reports a the number of items or rows in the oplog (e.g. 0). db.getReplicationInfo.timeDiff Returns the difference between the rst and last operation in the oplog, represented in seconds. Only present if there are entires in the oplog.
979
db.getReplicationInfo.timeDiffHours Returns the difference between the rst and last operation in the oplog, rounded and represented in hours. Only present if there are entires in the oplog. db.getReplicationInfo.tFirst Returns a time stamp for the rst (i.e. earliest) operation in the oplog. Compare this value to the last write operation issued against the server. Only present if there are entires in the oplog. db.getReplicationInfo.tLast Returns a time stamp for the last (i.e. latest) operation in the oplog. Compare this value to the last write operation issued against the server. Only present if there are entires in the oplog. db.getReplicationInfo.now Returns a time stamp that reects reecting the current time. The shell process generates this value, and the datum may differ slightly from the server time if youre connecting from a remote host as a result. Equivalent to Date(). Only present if there are entires in the oplog.
db.getSiblingDB()
db.getSiblingDB() Used to return another database without modifying the db variable in the shell environment.
db.help()
db.help() Returns Text output listing common methods on the db object.

db.hostInfo()
db.hostInfo() New in version 2.2. Returns A document with information about the underlying system that the mongod (page 1021) or mongos (page 1032) runs on. Some of the returned elds are only included on some platforms. db.hostInfo() (page 980) provides a helper in the mongo (page 1036) shell around the hostInfo (page 880) The output of db.hostInfo() (page 980) on a Linux system will resemble the following:
{ "system" : { "currentTime" : ISODate("<timestamp>"), "hostname" : "<hostname>", "cpuAddrSize" : <number>, "memSizeMB" : <number>, "numCores" : <number>, "cpuArch" : "<identifier>", "numaEnabled" : <boolean> }, "os" : {
980
"type" : "<string>", "name" : "<string>", "version" : "<string>" }, "extra" : { "versionString" : "<string>", "libcVersion" : "<string>", "kernelVersion" : "<string>", "cpuFrequencyMHz" : "<string>", "cpuFeatures" : "<string>", "pageSize" : <number>, "numPages" : <number>, "maxOpenFiles" : <number> }, "ok" : <return> }
See hostInfo (page 881) for full documentation of the output of db.hostInfo() (page 980).
db.isMaster()
db.isMaster() Returns A document that describes the role of the mongod (page 1021) instance. If the mongod (page 1021) is a member of a replica set, then the ismaster (page 839) and secondary (page 839) elds report if the instance is the primary or if it is a secondary member of the replica set. See also: isMaster (page 838) for the complete documentation of the output of isMaster() (page 981).
db.killOp()
db.killOp(opid) Parameters opid Specify an operation ID. Terminates the specied operation. Use db.currentOp() (page 971) to nd operations and their corresponding ids. See db.currentOp() (page 971) for full documentation of the output of db.currentOp() (page 971). Warning: Terminate running operations with extreme caution. Only use db.killOp() (page 981) to terminate operations initiated by clients and do not terminate internal database operations.
db.listCommands()
db.listCommands() Provides a list of all database commands. See the Database Commands (page 807) document for a more extensive index of these options.
981
db.loadServerScripts()
db.loadServerScripts() db.loadServerScripts() (page 982) loads all scripts in the system.js collection for the current database into the mongo (page 1036) shell session. Documents in the system.js collection have the following prototype form:
{ _id : "<name>" , value : <function> } }
The documents in the system.js collection provide functions that your applications can use in any JavaScript context with MongoDB in this database. These contexts include $where (page 775) clauses and mapReduce (page 814) operations.
db.logout()
db.logout() Ends the current authentication session. This function has no effect if the current session is not authenticated. Note: If youre not logged in and using authentication, db.logout() (page 982) has no effect. Changed in version 2.4: Because MongoDB now allows users dened in one database to have privileges on another database, you must call db.logout() (page 982) while using the same database context that you authenticated to. If you authenticated to a database such as users or $external, you must issue db.logout() (page 982) against this database in order to successfully log out. Example Use the use <database-name> helper in the interactive mongo (page 1036) shell, or the following db.getSiblingDB() (page 980) in the interactive shell or in mongo (page 1036) shell scripts to change the db object:
db = db.getSiblingDB(<database-name>)
When you have set the database context and db object, you can use the db.logout() (page 982) to log out of database as in the following operation:
db.logout()
db.logout() (page 982) function provides a wrapper around the database command logout.
db.printCollectionStats()
db.printCollectionStats() Provides a wrapper around the db.collection.stats() (page 947) method. Returns statistics from every collection separated by three hyphen characters. Note: The db.printCollectionStats() (page 982) in the mongo (page 1036) shell does not return JSON . Use db.printCollectionStats() (page 982) for manual inspection, and db.collection.stats() (page 947) in scripts. See also:
982
collStats (page 873)

db.printReplicationInfo()
db.printReplicationInfo() Provides a formatted report of the status of a replica set from the perspective of the primary set member. See the replSetGetStatus (page 840) for more information regarding the contents of this output. Note: The db.printReplicationInfo() (page 983) in the mongo (page 1036) shell does not return JSON . Use db.printReplicationInfo() (page 983) for manual inspection, and rs.status() (page 990) in scripts.
db.printShardingStatus()
db.printShardingStatus() Prints a formatted report of the sharding conguration and the information regarding existing chunks in a sharded cluster. Only use db.printShardingStatus() (page 983) when connected to a mongos (page 1032) instance. Parameters verbose (boolean) Optional. If true, the method displays details of the document distribution across chunks when you have 20 or more chunks. See sh.status() (page 1002) for details of the output. Note: The db.printShardingStatus() (page 983) in the mongo (page 1036) shell does not return JSON . Use db.printShardingStatus() (page 983) for manual inspection, and Cong Database (page 545) in scripts. See also: sh.status() (page 1002)
db.printSlaveReplicationInfo()
db.printSlaveReplicationInfo() Provides a formatted report of the status of a replica set from the perspective of the secondary set member. See the replSetGetStatus (page 840) for more information regarding the contents of this output. Note: The db.printSlaveReplicationInfo() (page 983) in the mongo (page 1036) shell does not return JSON . Use db.printSlaveReplicationInfo() (page 983) for manual inspection, and rs.status() (page 990) in scripts.
db.removeUser()
db.removeUser(username) Parameters username Specify a database username.
983
Removes the specied username from the database.

db.repairDatabase()
db.repairDatabase() Warning: In general, if you have an intact copy of your data, such as would exist on a very recent backup or an intact member of a replica set, do not use repairDatabase (page 868) or related options like db.repairDatabase() (page 984) in the mongo (page 1036) shell or mongod --repair (page 1026). Restore from an intact copy of your data.
Note: When using journaling, there is almost never any need to run repairDatabase (page 868). In the event of an unclean shutdown, the server will be able restore the data les to a pristine state automatically. db.repairDatabase() (page 984) provides a wrapper around the database command repairDatabase (page 868), and has the same effect as the run-time option mongod --repair (page 1026) option, limited to only the current database. See repairDatabase (page 868) for full documentation.
db.resetError()
db.resetError() Deprecated since version 1.6. Resets the error message returned by db.getPrevError (page 979) or getPrevError (page 832). Provides a wrapper around the resetError (page 832) command.
db.runCommand()
db.runCommand(command) Parameters command (string) Species a database command in the form of a document. command When specifying a command (page 807) as a string, db.runCommand() (page 984) transforms the command into the form { command: 1 }. Provides a helper to run specied database commands (page 807). This is the preferred method to issue database commands, as it provides a consistent interface between the shell and drivers.
db.serverBuildInfo()
db.serverBuildInfo() Provides a wrapper around the buildInfo (page 871) database command. buildInfo (page 871) returns a document that contains an overview of parameters used to compile this mongod (page 1021) instance.
db.serverStatus()
db.serverStatus() Returns a document that provides an overview of the database processs state. This command provides a wrapper around the database command serverStatus (page 889).
984
Changed in version 2.4: In 2.4 you can dynamically suppress portions of the db.serverStatus() (page 984) output, or include suppressed sections in a document passed to the db.serverStatus() (page 984) method, as in the following example:
db.serverStatus( { repl: 0, indexCounters: 0, locks: 0 } ) db.serverStatus( { workingSet: 1, metrics: 0, locks: 0 } )
db.serverStatus() (page 984) includes all elds by default, except workingSet (page 902), by default. Note: You may only dynamically include top-level elds from the serverStatus (page 889) document that are not included by default. You can exclude any eld that db.serverStatus() (page 984) includes by default. See also: serverStatus (page 889) for complete documentation of the output of this function.
db.setProlingLevel()
db.setProfilingLevel(level[, slowms ]) Parameters level Species a proling level, see list of possible values below. slowms Optionally modify the threshold for the prole to consider a query or operation slow. Modies the current database proler level. This allows administrators to capture data regarding performance. The database proling system can impact performance and can allow the server to write the contents of queries to the log, which might have information security implications for your deployment. The following proling levels are available: Level 0 1 2 Setting Off. No proling. On. Only includes slow operations. On. Includes all operations.
Also congure the slowms (page 1085) option to set the threshold for the proler to consider a query slow. Specify this value in milliseconds to override the default. This command provides a wrapper around the database command profile (page 888). mongod (page 1021) writes the output of the database proler to the system.profile collection. mongod (page 1021) prints information about queries that take longer than the slowms (page 1085) to the log even when the database proler is not active. Note: The database cannot be locked with db.fsyncLock() (page 977) while proling is enabled. You must disable proling before locking the database with db.fsyncLock() (page 977). Disable proling using db.setProfilingLevel() (page 985) as follows in the mongo (page 1036) shell:
db.shutdownServer()
db.shutdownServer() Shuts down the current mongod (page 1021) or mongos (page 1032) process cleanly and safely. 63.3. mongo Shell Methods 985
This operation fails when the current database is not the admin database. This command provides a wrapper around the shutdown (page 869).
db.stats()
db.stats(scale) Parameters scale Optional. Species the scale to deliver results. Unless specied, this command returns all data in bytes. Returns A document that contains statistics reecting the database systems state. This function provides a wrapper around the database command dbStats (page 877). To convert the returned values to kilobytes, use the scale argument:
db.stats(1024) The :doc:dbStats command </reference/command/dbStats> document describes the fields in the :method:db.stats() output.
Note: The scale factor rounds values to whole numbers. This can produce unpredictable and unexpected results in some situations.
db.version()
db.version() Returns The version of the mongod (page 1021) instance.
986
63.3.4 Replication
Replication Methods Name rs.add() (page 987) rs.addArb() (page 988) rs.conf() (page 988) rs.freeze() (page 988) rs.help() (page 989) rs.initiate() (page 989) rs.reconfig() (page 989) rs.remove() (page 990) rs.slaveOk() (page 990) rs.status() (page 990) rs.stepDown() (page 991) rs.syncFrom (page 991)
rs.add()
Description Adds a member to a replica set. Adds an arbiter to a replica set. Returns the replica set conguration document. Prevents the current member from seeking election as primary for a period of time. Returns basic help text for replica set functions. Initializes a new replica set. Re-congures a replica set by applying a new replica set conguration object. Remove a member from a replica set. Sets the slaveOk property for the current connection. Deprecated. Use readPref() (page 963) and Mongo.setReadPref() (page 1008) to set read preference. Returns a document with information about the state of the replica set. Causes the current primary to become a secondary which forces an election. Sets the member that this replica set member will sync from, overriding the default sync target selection logic.
Denition rs.add(host, arbiterOnly) Adds a member to a replica set. param string,document host Either a string or a document. If a string, species a host (and optionally port-number) for a new host member for the replica set; MongoDB will add this host with the default conguration. If a document, species any attributes about a member of a replica set. param boolean arbiterOnly If true, this host is an arbiter. If the second argument evaluates to true, as is the case with some documents, then this instance will become an arbiter. You may specify new hosts in one of two ways: 1.as a hostname with an optional port number to use the default conguration as in the Add a Member to an Existing Replica Set (page 425) example. 2.as a conguration document, as in the Add a Member to an Existing Replica Set (Alternate Procedure) (page 426) example. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. rs.add() (page 987) provides a wrapper around some of the functionality of the replSetReconfig (page 844) database command and the corresponding shell helper rs.reconfig() (page 989). See the 63.3. mongo Shell Methods 987
Replica Set Conguration (page 463) document for full documentation of all replica set conguration options. Example To add a mongod (page 1021) accessible on the default port 27017 running on the host mongodb3.example.net, use the following rs.add() (page 987) invocation:
rs.add(mongodb3.example.net:27017)
If mongodb3.example.net is an arbiter, use the following form:

rs.add(mongodb3.example.net:27017, true)
To add mongodb3.example.net as a secondary-only (page 388) member of set, use the following form of rs.add() (page 987):
rs.add( { "_id": "3", "host": "mongodbd3.example.net:27017", "priority": 0 } )
Replace, 3 with the next unused _id value in the replica set. See rs.conf() (page 988) to see the existing _id values in the replica set conguration document. See the Replica Set Conguration (page 463) and Replica Set Administration (page 419) documents for more information.
rs.addArb()
rs.addArb(hostname) Parameters host (string) Species a host (and optionally port-number) for an arbiter member for the replica set. Adds a new arbiter to an existing replica set. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds.
rs.conf()
rs.conf() Returns a document that contains the current replica set conguration object. rs.config() rs.config() (page 988) is an alias of rs.conf() (page 988).
rs.freeze()
rs.freeze(seconds) Parameters seconds (init) Specify the duration of this operation. Forces the current node to become ineligible to become primary for the period specied. rs.freeze() (page 988) provides a wrapper around the database command replSetFreeze (page 840).
988
rs.help()
rs.help() Returns a basic help text for all of the replication (page 387) related shell functions.
rs.initiate()
rs.initiate(conguration) Parameters conguration Optional. A document that species the conguration of a replica set. If not specied, MongoDB will use a default conguration. Initiates a replica set. Optionally takes a conguration argument in the form of a document that holds the conguration of a replica set. Consider the following model of the most basic conguration for a 3-member replica set:
This function provides a wrapper around the replSetInitiate (page 842) database command.
rs.recong()
rs.reconfig(conguration[, force ]) Parameters conguration A document that species the conguration of a replica set. force Optional. Specify { force: true } as the force parameter to force the replica set to accept the new conguration even if a majority of the members are not accessible. Use with caution, as this can lead to rollback situations. Initializes a new replica set conguration. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. rs.reconfig() (page 989) provides a wrapper around the replSetReconfig (page 844) database command. rs.reconfig() (page 989) overwrites the existing replica set conguration. Retrieve the current conguration object with rs.conf() (page 988), modify the conguration as needed and then use rs.reconfig() (page 989) to submit the modied conguration object. To recongure a replica set, use the following sequence of operations:
conf = rs.conf() // modify conf to change configuration rs.reconfig(conf)
989
If you want to force the reconguration if a majority of the set isnt connected to the current member, or youre issuing the command against a secondary, use the following form:
conf = rs.conf() // modify conf to change configuration rs.reconfig(conf, { force: true } )
Warning: Forcing a rs.reconfig() (page 989) can lead to rollback situations and other difcult to recover from situations. Exercise caution when using this option. See also: Replica Set Conguration (page 463) and Replica Set Administration (page 419).
rs.remove()
rs.remove(hostname) Parameters hostname Specify one of the existing hosts to remove from the current replica set. Removes the node described by the hostname parameter from the current replica set. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. Note: Before running the rs.remove() (page 990) operation, you must shut down the replica set member that youre removing. Changed in version 2.2: This procedure is no longer required when using rs.remove() (page 990), but it remains good practice.
rs.slaveOk()
rs.slaveOk() Provides a shorthand for the following operation:

This allows the current connection to allow read operations to run on secondary nodes. See the readPref() (page 963) method for more ne-grained control over read preference (page 402) in the mongo (page 1036) shell.
rs.status()
rs.status() Returns A document with status information. This output reects the current status of the replica set, using data derived from the heartbeat packets sent by the other members of the replica set.
990
This method provides a wrapper around the replSetGetStatus (page 840) database command.
rs.stepDown()
rs.stepDown(seconds) Parameters seconds (init) Specify the duration of this operation. If not specied the command uses the default value of 60 seconds. Returns disconnects shell. Forces the current replica set member to step down as primary and then attempt to avoid election as primary for the designated number of seconds. Produces an error if the current node is not primary. This function will disconnect the shell briey and forces a reconnection as the replica set renegotiates which node will be primary. As a result, the shell will display an error even if this command succeeds. rs.stepDown() (page 991) provides a wrapper around the database command replSetStepDown (page 844).
rs.syncFrom()
rs.syncFrom() New in version 2.2. Provides a wrapper around the replSetSyncFrom (page 845), which allows administrators to congure the member of a replica set that the current member will pull data from. Specify the name of the member you want to replicate from in the form of [hostname]:[port]. See replSetSyncFrom (page 845) for more details.
991
992
63.3.5 Sharding
Sharding Methods Name Description sh._adminCommand Runs a database command against the admin database, like db.runCommand() (page 994) (page 984), but can conrm that it is issued against a mongos (page 1032). sh._checkFullName() Tests a namespace to determine if its well formed. (page 994) sh._checkMongos() Tests to see if the mongo (page 1036) shell is connected to a mongos (page 1032) (page 994) instance. sh._lastMigration() Reports on the last chunk migration. (page 994) sh.addShard() Adds a shard to a sharded cluster. (page 995) sh.addShardTag() Associates a shard with a tag, to support tag aware sharding (page 532). (page 996) sh.addTagRange() Associates range of shard keys with a shard tag, to support tag aware sharding (page 532). sh.disableBalancing() Deactivates the cluster balancer process which manages chunk migrations between (page 997) shards. sh.enableBalancing() Activates the cluster balancer process if previously disabled using (page 997) sh.disableBalancing() (page 997). sh.enableSharding() Enables sharding on a specic database. (page 998) sh.getBalancerHost() Returns the name of a mongos (page 1032) thats responsible for the balancer (page 998) process. sh.getBalancerState() Returns a boolean to report if the balancer is currently enabled. (page 998) sh.help() (page 999) Returns help text for the sh methods. sh.isBalancerRunning() Returns a boolean to report if the balancer process is currently migrating chunks. (page 999) sh.moveChunk() Migrates a chunk in a sharded cluster. (page 999) sh.removeShardTag() Removes the association between a shard and a shard tag shard tag. (page 1000) sh.setBalancerState() Enables or disables the balancer which migrates chunks between shards. (page 1000) sh.shardCollection() Enables sharding for a collection. (page 1000) sh.splitAt() Divides an existing chunk into two chunks using a specic value of the shard key as (page 1001) the dividing point. sh.splitFind() Divides an existing chunk that contains a document matching a query into two (page 1001) approximately equal chunks. sh.startBalancer() Enables the balancer and waits for balancing to start. (page 1002) sh.status() Reports on the status of a sharded cluster, as db.printShardingStatus() (page 1002) (page 983). sh.stopBalancer() Disables the balancer and waits for any in progress balancing rounds to complete. (page 1004) sh.waitForBalancer() Internal. Waits for the balancer state to change. (page 1005) sh.waitForBalancerOff() Internal. Waits until the balancer stops running. (page 1005) sh.waitForDLock() Internal. Waits for a specied distributed sharded cluster lock. 63.3. 993 (pagemongo 1006) Shell Methods sh.waitForPingChange() Internal. Waits for a change in ping state from one of the mongos (page 1032) in the (page 1006) sharded cluster.
sh._adminCommand()
sh._adminCommand(cmd, checkMongos) Parameters dbcommand (string) A database command to run against the admin database. checkMongos (Boolean) Verify whether or not the shell is connected to a mongos (page 1032) instance. The sh._adminCommand (page 994) method runs a database command against the admin database of a mongos (page 1032) instance. See also: db.runCommand() (page 984)
sh._checkFullName()
sh._checkFullName(namespace) Parameters namespace (string) Specify a complete namespace. Throws name needs to be fully qualied <db>.<collection> The sh._checkFullName() (page 994) method veries that a namespace name is well-formed. If the name has a period . then the sh._checkFullName() (page 994) method exits, otherwise it throws an error.
sh._checkMongos()
sh._checkMongos() Returns nothing Throws not connected to a mongos The sh._checkMongos() (page 994) method throws an error message if the mongo (page 1036) shell is not connected to a mongos (page 1032) instance. Otherwise it exits (no return document or return code).
sh._lastMigration()
sh._lastMigration(namespace) Parameters namespace (string) The name of a database or collection within the current database. Returns A document with elds detailing the most recent migration in the specied namespace. sh._lastMigration() (page 994) returns a document with details about the last migration performed on the database or collection you specify. Document details: Fields _id (string) The id of the migration task
994
server (string) The name of the server clientAddr (string) The IP address and port number of the server. time (ISODate) The time of the last migration. what (string) The specic type of migration. ns (string) The complete namespace of the collection affected by the migration. details (document) A document containing details about the migrated chunk. Includes min and max sub-documents with the bounds of the migrated chunk.
sh.addShard()
sh.addShard(host) Parameters host (string) Specify the hostname of a database instance or a replica set conguration. Use this method to add a database instance or replica set to a sharded cluster. This method must be run on a mongos (page 1032) instance. The host parameter can be in any of the following forms:
[hostname] [hostname]:[port] [set]/[hostname] [set]/[hostname],[hostname]:port
You can specify shards using the hostname, or a hostname and port combination if the shard is running on a non-standard port. Warning: Do not use localhost for the hostname unless your conguration server is also running on localhost. The optimal conguration is to deploy shards across replica sets. To add a shard on a replica set you must specify the name of the replica set and the hostname of at least one member of the replica set. You must specify at least one member of the set, but can specify all members in the set or another subset if desired. sh.addShard() (page 995) takes the following form: If you specify additional hostnames, all must be members of the same replica set.
sh.addShard("set-name/seed-hostname")
Example
sh.addShard("repl0/mongodb3.example.net:27327")
The sh.addShard() (page 995) method is a helper for the addShard (page 846) command. addShard (page 846) command has additional options which are not available with this helper. See also: addShard (page 846) Sharded Cluster Administration (page 503) Add Shards to a Cluster (page 510) Remove Shards from an Existing Sharded Cluster (page 530)
The
995
sh.addShardTag()
sh.addShardTag(shard, tag) New in version 2.2. Parameters shard (string) Species the name of the shard that you want to give a specic tag. tag (string) Species the name of the tag that you want to add to the shard. sh.addShardTag() (page 996) associates a shard with a tag or identier. MongoDB uses these identiers to direct chunks that fall within a tagged range to specic shards. sh.addTagRange() associates chunk ranges with tag ranges. Always issue sh.addShardTag() (page 996) when connected to a mongos (page 1032) instance. Example The following example adds three tags, NYC, LAX, and NRT, to three shards:
See also: sh.addTagRange() and sh.removeShardTag() (page 1000)

sh.addShardTag()
sh.addShardTag(shard, tag) New in version 2.2. Parameters shard (string) Species the name of the shard that you want to give a specic tag. tag (string) Species the name of the tag that you want to add to the shard. sh.addShardTag() (page 996) associates a shard with a tag or identier. MongoDB uses these identiers to direct chunks that fall within a tagged range to specic shards. sh.addTagRange() associates chunk ranges with tag ranges. Always issue sh.addShardTag() (page 996) when connected to a mongos (page 1032) instance. Example The following example adds three tags, NYC, LAX, and NRT, to three shards:
See also:
996
sh.addTagRange() and sh.removeShardTag() (page 1000)

sh.disableBalancing()
sh.disableBalancing(collection) Parameters collection (string) The name of a collection. sh.disableBalancing() (page 997) disables the balancer for the specied sharded collection. See also: sh.enableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.setBalancerState() (page 1000) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005)
sh.enableBalancing()
sh.enableBalancing(collection) Parameters collection (string) The name of a collection. sh.enableBalancing() (page 997) enables the balancer for the specied sharded collection. See also: sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.setBalancerState() (page 1000) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005)
997
sh.enableSharding()
sh.enableSharding(database) Parameters database (string) Specify a database name to shard. Enables sharding on the specied database. This does not automatically shard any collections, but makes it possible to begin sharding collections using sh.shardCollection() (page 1000). See also: sh.shardCollection() (page 1000)
sh.getBalancerHost()
sh.getBalancerHost() Returns String in form hostname:port sh.getBalancerHost() (page 998) returns the name of the server that is running the balancer. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.setBalancerState() (page 1000) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005)
sh.getBalancerState()
sh.getBalancerState() Returns boolean sh.getBalancerState() (page 998) returns true when the balancer is enabled and false if the balancer is disabled. This does not reect the current state of balancing operations: use sh.isBalancerRunning() (page 999) to check the balancers current state. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.isBalancerRunning() (page 999) sh.setBalancerState() (page 1000)
998
sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005)
sh.help()
sh.help() Returns a basic help text for all sharding related shell functions.
sh.isBalancerRunning()
sh.isBalancerRunning() Returns boolean Returns true if the balancer process is currently running and migrating chunks and false if the balancer process is not running. Use sh.getBalancerState() (page 998) to determine if the balancer is enabled or disabled. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.setBalancerState() (page 1000) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005)
sh.moveChunk()
sh.moveChunk(collection, query, destination) Parameters collection (string) Species the sharded collection containing the chunk to migrate. query (document) A document that species an equality match on the shard key, which selects the chunk to move. destination (string) Species the name of the shard that you wish to move the designated chunk to. Moves the chunk containing the document specied by the query to the shard described by destination. This method provides a wrapper around the moveChunk (page 849). In most circumstances, allow the balancer to automatically migrate chunks, and avoid calling sh.moveChunk() (page 999) directly. See also: 63.3. mongo Shell Methods 999
moveChunk (page 849), sh.splitAt() (page 1001), sh.splitFind() (page 1001), Sharding (page 483), and chunk migration (page 499).
sh.removeShardTag()
sh.removeShardTag(shard, tag) New in version 2.2. Parameters shard (string) Species the name of the shard that you want to remove a tag from. tag (string) Species the name of the tag that you want to remove from the shard. Removes the association between a tag and a shard. Always issue sh.removeShardTag() (page 1000) when connected to a mongos (page 1032) instance. See also: sh.addShardTag() (page 996), sh.addTagRange()
sh.setBalancerState()
sh.setBalancerState(state) Parameters state (boolean) true enables the balancer if disabled, and false disables the balancer. Enables or disables the balancer. Use sh.getBalancerState() (page 998) to determine if the balancer is currently enabled or disabled and sh.isBalancerRunning() (page 999) to check its current state. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005)
sh.shardCollection()
sh.shardCollection(namespace, key, unique) Parameters namespace (string) The namespace of the collection to shard. key (document) A document containing a shard key that the sharding system uses to partition and distribute objects among the shards. 1000 Chapter 63. MongoDB Interface
unique (boolean) When true, the unique option ensures that the underlying index enforces a unique constraint. Hashed shard keys do not support unique constraints. Shards the named collection, according to the specied shard key. Specify shard keys in the form of a document. Shard keys may refer to a single document eld, or more typically several document elds to form a compound shard key. New in version 2.4: Use the form {field: may not be compound indexes. "hashed"} to create a hashed shard key. Hashed shard keys
Warning: MongoDB provides no method to deactivate sharding for a collection after calling shardCollection (page 851). Additionally, after shardCollection (page 851), you cannot change shard keys or modify the value of any eld used in your shard key index. See also: shardCollection (page 851) for additional options, Sharding (page 483), Sharded Cluster Overview (page 485) for an overview of sharding with MongoDB and Deploy a Sharded Cluster (page 503) for a tutorial. Also review Shard Keys (page 485) regarding choosing a shard key.
sh.splitAt()
sh.splitAt(namespace, query) Parameters namespace (string) Specify the namespace (i.e. <database>.<collection>) of the sharded collection that contains the chunk to split. query (document) Specify a query to identify a document in a specic chunk. Typically specify the shard key for a document as the query. Splits the chunk containing the document specied by the query as if that document were at the middle of the collection, even if the specied document is not the actual median of the collection. Use this command to manually split chunks unevenly. Use the sh.splitFind() (page 1001) function to split a chunk at the actual median. In most circumstances, you should leave chunk splitting to the automated processes within MongoDB. However, when initially deploying a sharded cluster it is necessary to perform some measure of pre-splitting using manual methods including sh.splitAt() (page 1001).
sh.splitFind()
sh.splitFind(namespace, query) Parameters namespace (string) Specify the namespace (i.e. <database>.<collection>) of the sharded collection that contains the chunk to split. query Specify a query to identify a document in a specic chunk. Typically specify the shard key for a document as the query. Splits the chunk containing the document specied by the query at its median point, creating two roughly equal chunks. Use sh.splitAt() (page 1001) to split a collection in a specic point.
1001
In most circumstances, you should leave chunk splitting to the automated processes. However, when initially deploying a sharded cluster it is necessary to perform some measure of pre-splitting using manual methods including sh.splitFind() (page 1001).
sh.startBalancer()
sh.startBalancer(timeout, interval) Parameters timeout (integer) Milliseconds to wait. interval (integer) Milliseconds to sleep each cycle of waiting. The sh.startBalancer() (page 1002) enables the balancer in a sharded cluster and waits for balancing to initiate. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.setBalancerState() (page 1000) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005)
sh.status()
Denition sh.status() Prints a formatted report of the sharding conguration and the information regarding existing chunks in a sharded cluster. The default behavior suppresses the detailed chunk information if the total number of chunks is greater than or equal to 20. Parameters verbose (boolean) Optional. If true, the method displays details of the document distribution across chunks when you have 20 or more chunks. See also: db.printShardingStatus() (page 983) Example Output The Sharding Version (page 1003) section displays information on the cong database:
--- Sharding Status --sharding version: { "_id" : <num>, "version" : <num>, "minCompatibleVersion" : <num>,
1002
"currentVersion" : <num>, "clusterId" : <ObjectId> }
The Shards (page 1004) section lists information on the shard(s). For each shard, the section displays the name, host, and the associated tags, if any.
shards: { "_id" : <shard name1>, "host" : <string>, "tags" : [ <string> ... ] } { "_id" : <shard name2>, "host" : <string>, "tags" : [ <string> ... ] } ...
The Databases (page 1004) section lists information on the database(s). For each database, the section displays the name, whether the database has sharding enabled, and the primary shard for the database.
databases: { "_id" : <dbname1>, "partitioned" : <boolean>, "primary" : <string> } { "_id" : <dbname2>, "partitioned" : <boolean>, "primary" : <string> } ...
The Sharded Collection (page 1004) section provides information on the sharding details for sharded collection(s). For each sharded collection, the section displays the shard key, the number of chunks per shard(s), the distribution of documents across chunks 3 , and the tag information, if any, for shard key range(s).
<dbname>.<collection> shard key: { <shard key> : <1 or hashed> } chunks: <shard name1> <number of chunks> <shard name2> <number of chunks> ... { <shard key>: <min range1> } -->> { <shard key> : <max range1> } on : <shard name> <last modified { <shard key>: <min range2> } -->> { <shard key> : <max range2> } on : <shard name> <last modified ... tag: <tag1> { <shard key> : <min range1> } -->> { <shard key> : <max range1> } ...
Output Sharding Version sh.status.sharding-version._id The _id (page 1003) is an identier for the version details.
3 The sharded collection section, by default, displays the chunk information if the total number of chunks is less than 20. To display the information when you have 20 or more chunks, call the the sh.status() (page 1002) methods with the verbose parameter set to true, i.e. sh.status(true).
1003
sh.status.sharding-version.version The version (page 1004) is the version of the cong server for the sharded cluster. sh.status.sharding-version.minCompatibleVersion The minCompatibleVersion (page 1004) is the minimum compatible version of the cong server. sh.status.sharding-version.currentVersion The currentVersion (page 1004) is the current version of the cong server. sh.status.sharding-version.clusterId The clusterId (page 1004) is the identication for the sharded cluster. Shards sh.status.shards._id The _id (page 1004) displays the name of the shard. sh.status.shards.host The host (page 1004) displays the host location of the shard. sh.status.shards.tags The tags (page 1004) displays all the tags for the shard. The eld only displays if the shard has tags. Databases sh.status.databases._id The _id (page 1004) displays the name of the database. sh.status.databases.partitioned The partitioned (page 1004) displays whether the database has sharding enabled. If true, the database has sharding enabled. sh.status.databases.primary The primary (page 1004) displays the primary shard for the database. Sharded Collection sh.status.databases.shard-key The shard-key (page 1004) displays the shard key specication document. sh.status.databases.chunks The chunks (page 1004) lists all the shards and the number of chunks that reside on each shard. sh.status.databases.chunk-details The chunk-details (page 1004) lists the details of the chunks 1 : The range of shard key values that dene the chunk, The shard where the chunk resides, and The last modied timestamp for the chunk. sh.status.databases.tag The tag (page 1004) lists the details of the tags associated with a range of shard key values.
sh.stopBalancer()
sh.stopBalancer(timeout, interval) Parameters timeout (integer) Milliseconds to wait. interval (integer) Milliseconds to sleep each cycle of waiting. 1004 Chapter 63. MongoDB Interface
The sh.stopBalancer() (page 1004) disables the balancer in a sharded cluster and waits for balancing to complete. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.setBalancerState() (page 1000) sh.startBalancer() (page 1002) sh.waitForBalancer() (page 1005) sh.waitForBalancerOff() (page 1005)
sh.waitForBalancer()
sh.waitForBalancer(onOrNot, timeout, interval) Parameters onOrNot (Boolean) Whether to wait for the lock to be on (true) or off (false). timeout (integer) Milliseconds to wait. interval (integer) Milliseconds to sleep. sh.waitForBalancer() (page 1005) is an internal method that waits for a change in the state of the balancer. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.setBalancerState() (page 1000) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancerOff() (page 1005)
sh.waitForBalancerOff()
sh.waitForBalancerOff() Parameters timeout (integer) Milliseconds to wait.
1005
interval (integer) Milliseconds to sleep. sh.waitForBalancerOff() (page 1005) is an internal method that waits until the balancer is not running. See also: sh.enableBalancing() (page 997) sh.disableBalancing() (page 997) sh.getBalancerHost() (page 998) sh.getBalancerState() (page 998) sh.isBalancerRunning() (page 999) sh.setBalancerState() (page 1000) sh.startBalancer() (page 1002) sh.stopBalancer() (page 1004) sh.waitForBalancer() (page 1005)
sh.waitForDLock()
sh.waitForDLock(lockId, onOrNot, timeout, interval) Parameters lockId (string) The name of the distributed lock. onOrNot (Boolean) Optional, whether to wait for the lock to be on (true) or off (false). timeout (integer) Milliseconds to wait. interval (integer) Milliseconds to sleep in each waiting cycle. sh.waitForDLock() (page 1006) is an internal method that waits until the specied distributed lock is changes state.
sh.waitForPingChange()
sh.waitForPingChange(activepings, timeout, interval) Parameters activepings (array) An array of active pings from the config.mongos collection. timeout (integer) Milliseconds to wait for a change in ping state. interval (integer) Milliseconds to sleep in each waiting cycle. sh.waitForPingChange() (page 1006) waits for a change in ping state of the one of the activepings.
1006
63.3.6 Connection
Connection Methods Name Mongo() (page 1007) Mongo.getDB() (page 1007) Mongo.getReadPrefMode() (page 1008) Mongo.getReadPrefTagSet() (page 1008) Mongo.setReadPref() (page 1008) Mongo.setSlaveOk() (page 1009) connect() Description Creates a new connection object. Returns a database object. Returns the current read preference mode for the MongoDB connection. Returns the read preference tag set for the MongoDB connection. Sets the read preference for the MongoDB connection. Allows operations on the current connection to read from secondary members. Connects to a MongoDB instance and to a specied database on that instance.
Mongo()
Mongo() JavaScript constructor to instantiate a database connection from the mongo (page 1036) shell or from a JavaScript le. Parameters host (string) Optional. Either in the form of <host> or <host><:port>. Pass the <host> parameter to the constructor to instantiate a connection to the <host> and the default port. Pass the <host><:port> parameter to the constructor to instantiate a connection to the <host> and the <port>. Use the constructor without a parameter to instantiate a connection to the localhost interface on the default port. See also: Mongo.getDB() (page 1007) and db.getMongo() (page 978).
Mongo.getDB()
Mongo.getDB(<database>) Mongo.getDB() (page 1007) provides access to database objects from the mongo (page 1036) shell or from a JavaScript le. Parameters database (string) The name of the database to access. The following example instantiates a new connection to the MongoDB instance running on the localhost interface and returns a reference to "myDatabase":
db = new Mongo().getDB("myDatabase");
See also: Mongo() (page 1007) and connect() (page 1010) 63.3. mongo Shell Methods 1007
Mongo.getReadPrefMode()
Mongo.getReadPrefMode() Returns The current read preference mode for the Mongo() (page 978) connection object. See Read Preference (page 402) for an introduction to read preferences in MongoDB. Use getReadPrefMode() (page 1008) to return the current read preference mode, as in the following example:
db.getMongo().getReadPrefMode()
Use the following operation to return and print the current read preference mode:
print(db.getMongo().getReadPrefMode());
This operation will return one of the following read preference modes: primary (page 403) primaryPreferred (page 403) secondary (page 403) secondaryPreferred (page 404) nearest (page 404) See also: Read Preference (page 402), cursor.readPref() (page 963), Mongo.setReadPref() (page 1008), and Mongo.getReadPrefTagSet() (page 1008).
Mongo.getReadPrefTagSet()
Mongo.getReadPrefTagSet() Returns The current read preference tag set for the Mongo() (page 978) connection object. See Read Preference (page 402) for an introduction to read preferences and tag sets in MongoDB. Use getReadPrefTagSet() (page 1008) to return the current read preference tag set, as in the following example:
db.getMongo().getReadPrefTagSet()
Use the following operation to return and print the current read preference tag set:
printjson(db.getMongo().getReadPrefTagSet());
See also: Read Preference (page 402), cursor.readPref() (page 963), Mongo.setReadPref() (page 1008), and Mongo.getReadPrefTagSet() (page 1008).
Mongo.setReadPref()
Mongo.setReadPref(<mode>, <tagSet>) Parameters mode (string) Read preference mode.
1008
tagSet (array) Optional. A tag set . Call the setReadPref() (page 1008) method on a Mongo (page 978) connection object to control how the client will route all queries to members of the replica set. The mode string should be one of: primary (page 403) primaryPreferred (page 403) secondary (page 403) secondaryPreferred (page 404) nearest (page 404) See the tag sets (page 404) documentation for more information on using tag sets to provide custom read preference modes. To set a read preference mode in the mongo (page 1036) shell, use the following operation:
db.getMongo().setReadPref(primaryPreferred)
To set a read preference that uses a tag set, specify an array of tag sets as the second argument to Mongo.setReadPref() (page 1008), as in the following:
db.getMongo().setReadPref(primaryPreferred, [ { "dc": "east" } ] )
You can specify multiple tag sets, in order of preference, as in the following:
db.getMongo().setReadPref(secondaryPreferred, [ { "dc": "east", "use": "production" }, { "dc": "east", "use": "reporting" }, { "dc": "east" }, {} ] )
If the replica set cannot satisfy the rst tag set, the client will attempt to use the second read preference. Each tag set can contain zero or more eld/value tag pairs, with an empty document acting as a wildcard which matches a replica set member with any tag set or no tag set. Note: You must call Mongo.setReadPref() (page 1008) on the connection object before retrieving documents using that connection to use that read preference.
mongo.setSlaveOk()
Mongo.setSlaveOk() For the current session, this command permits read operations from non-master (i.e. slave or secondary) instances. Practically, use this method in the following form:
Indicates that eventually consistent read operations are acceptable for the current application. This function provides the same functionality as rs.slaveOk() (page 990). See the readPref() (page 963) method for more ne-grained control over read preference (page 402) in the mongo (page 1036) shell.
1009
connect()
connect(<hostname><:port>/<database>) The connect() method creates a connection to a MongoDB instance. (page 1007) object and its getDB() (page 1007) method in most cases.
However, use the Mongo()
connect() accepts a string <hostname><:port>/<database> parameter to connect to the MongoDB instance on the <hostname><:port> and return the reference to the database <database>. The following example instantiates a new connection to the MongoDB instance running on the localhost interface and returns a reference to myDatabase:
db = connect("localhost:27017/myDatabase")
See also: Mongo.getDB() (page 1007)
63.3.7 Subprocess
Subprocess Methods Name clearRawMongoProgramOutput() (page 1010) rawMongoProgramOutput() (page 1010) run() runMongoProgram() (page 1011) runProgram() (page 1011) startMongoProgram() stopMongoProgram() (page 1011) stopMongoProgramByPid() (page 1011) stopMongod() (page 1011) waitMongoProgramOnPort() (page 1011) waitProgram() (page 1011)
clearRawMongoProgramOutput()
Description For internal use. For internal use. For internal use. For internal use. For internal use. For internal use. For internal use. For internal use. For internal use. For internal use. For internal use.
clearRawMongoProgramOutput() For internal use.

rawMongoProgramOutput()
rawMongoProgramOutput() For internal use.

run()
run() For internal use.
1010
runMongoProgram()
runMongoProgram() For internal use.

runProgram()
runProgram() For internal use.

startMongoProgram()
_startMongoProgram() For internal use.

stopMongoProgram()
stopMongoProgram() For internal use.

stopMongoProgramByPid()
stopMongoProgramByPid() For internal use.

stopMongod()
stopMongod() For internal use.

waitMongoProgramOnPort()
waitMongoProgramOnPort() For internal use.

waitProgram()
waitProgram() For internal use.
1011
63.3.8 Native
Native Methods Name _isWindows() (page 1012) _rand() (page 1012) _srand() (page 1013) cat() cd() copyDbpath() (page 1013) fuzzFile() (page 1013) getHostName() (page 1013) getMemInfo() (page 1014) hostname() listFiles() (page 1014) load() ls() md5sumFile() (page 1015) mkdir() pwd() quit() removeFile() (page 1015) resetDbpath() (page 1015)
_isWindows()
Description Returns true if the shell runs on a Windows system; false if a Unix or Linux system. Returns a random number between 0 and 1. For internal use. Returns the contents of the specied le. Changes the current working directory to the specied path. Copies a local dbpath (page 1081). For internal use. For internal use to support testing. Returns the hostname of the system running the mongo (page 1036) shell. Returns a document that reports the amount of memory used by the shell. Returns the hostname of the system running the shell. Returns an array of documents that give the name and size of each object in the directory. Loads and runs a JavaScript le in the shell. Returns a list of the les in the current directory. The md5 hash of the specied le. Creates a directory at the specied path. Returns the current directory. Exits the current shell session. Removes the specied le from the local le system. Removes a local dbpath (page 1081). For internal use.
_isWindows() Returns boolean. Returns true if the mongo (page 1036) shell is running on a system that is Windows, or false if the shell is running on a Unix or Linux systems.
rand()
_rand() Returns A random number between 0 and 1. This function provides functionality similar to the Math.rand() function from the standard library.
1012
_srand()
_srand() For internal use.

cat()
cat(lename) Parameters lename (string) Specify a path and le name on the local le system. Returns the contents of the specied le. This function returns with output relative to the current shell session, and does not impact the server.
cd()
cd(path) Parameters path (string) Specify a path on the local le system. Changes the current working directory to the specied path. This function returns with output relative to the current shell session, and does not impact the server.
copyDbpath()
copyDbpath() For internal use.

fuzzFile()
fuzzFile(lename) Parameters lename (string) Specify a lename or path to a local le. Returns null For internal use.
getHostName()
getHostName() Returns The hostname of the system running the mongo (page 1036) shell process.
1013
getMemInfo()
getMemInfo() Returns a document with two elds that report the amount of memory used by the JavaScript shell process. The elds returned are resident and virtual.
hostname()
hostname() Returns The hostname of the system running the mongo (page 1036) shell process.
listFiles()
listFiles() Returns an array, containing one document per object in the directory. This function operates in the context of the mongo (page 1036) process. The included elds are: name Returns a string which contains the name of the object. isDirectory Returns true or false if the object is a directory. size Returns the size of the object in bytes. This eld is only present for les.
load()
load(<le>) Parameters le (string) Specify a path and le name containing JavaScript. This native function loads and runs a JavaScript le into the current shell environment. To run JavaScript with the mongo shell, you can either: use the --eval (page 1037) option when invoking the shell to evaluate a small amount of JavaScript code, or specify a le name with mongo (page 1038). mongo (page 1036) will execute the script and then exit. Add the --shell (page 1036) option to return to the shell after running the command. Specify les loaded with the load() function in relative terms to the current directory of the mongo (page 1036) shell session. Check the current directory using the pwd() function.
ls()
ls() Returns a list of the les in the current directory. This function returns with output relative to the current shell session, and does not impact the server.
1014
md5sumFile()
md5sumFile(lename) Parameters lename (string) a le name. Returns The md5 hash of the specied le. Note: The specied lename must refer to a le located on the system running the mongo (page 1036) shell.
mkdir()
mkdir(path) Parameters path (string) A path on the local lesystem. Creates a directory at the specied path. This command will create the entire path specied, if the enclosing directory or directories do not already exit. Equivalent to mkdir -p with BSD or GNU utilities.
pwd()
pwd() Returns the current directory. This function returns with output relative to the current shell session, and does not impact the server.
quit()
quit() Exits the current shell session.

removeFile()
removeFile(lename) Parameters lename (string) Specify a lename or path to a local le. Returns boolean. Removes the specied le from the local le system.
resetDbpath()
resetDbpath() For internal use.
1015
63.4 SQL to MongoDB Mapping Chart

In addition to the charts that follow, you might want to consider the Frequently Asked Questions (page 709) section for a selection of common questions about MongoDB.
63.4.1 Executables
The following table presents the MySQL/Oracle executables and the corresponding MongoDB executables. Database Server Database Client MySQL/Oracle mysqld/oracle mysql/sqlplus MongoDB mongod (page 1021) mongo (page 1036)
63.4.2 Terminology and Concepts

The following table presents the various SQL terminology and concepts and the corresponding MongoDB terminology and concepts. SQL Terms/Concepts database table row column index table joins primary key Specify any unique column or column combination as primary key. aggregation (e.g. group by) MongoDB Terms/Concepts database collection document or BSON document eld index embedded documents and linking primary key In MongoDB, the primary key is automatically set to the _id eld. aggregation framework See the SQL to Aggregation Framework Mapping Chart (page 307).
63.4.3 Examples
The following table presents the various SQL statements and the corresponding MongoDB statements. The examples in the table assume the following conditions: The SQL examples assume a table named users. The MongoDB examples assume a collection named users that contain documents of the following prototype:
{ _id: ObjectID("509a8fb2f3f4948bd2f983a0"), user_id: "abc123", age: 55, status: A }
Create and Alter The following table presents the various SQL statements related to table-level actions and the corresponding MongoDB statements.
1016
SQL Schema Statements CREATE TABLE users ( id MEDIUMINT NOT NULL AUTO_INCREMENT, user_id Varchar(30), age Number, status char(1), PRIMARY KEY (id) )
MongoDB Schema Statements Reference Implicitly created on rst insert See insert() (page 936) (page 936) operation. The primary and createCollection() key _id is automatically added if (page 970) for more information. _id eld is not specied. db.users.insert( { user_id: "abc123", age: 55, status: "A" } ) However, you can also explicitly create a collection: db.createCollection("users") Collections do not describe or en- See the Data Modeling Considerforce the structure of its documents; ations for MongoDB Applications i.e. there is no structural alteration at (page 233), update() (page 948), the collection level. and $set (page 788) for more inforHowever, at the document level, mation on changing the structure of update() (page 948) operations documents in a collection. can add elds to existing documents using the $set (page 788) operator. db.users.update( { }, { $set: { join_date: new Date() } }, { multi: true } ) Collections do not describe or enforce the structure of its documents; i.e. there is no structural alteration at the collection level. However, at the document level, update() (page 948) operations can remove elds from documents using the $unset (page 788) operator. db.users.update( { }, { $unset: { join_date: { multi: true } ) See Data Modeling Considerations for MongoDB Applications (page 233), update() (page 948), and $unset (page 788) for more information on changing the structure of documents in a collection.
ALTER TABLE users ADD join_date DATETIME
ALTER TABLE users DROP COLUMN join_date
"" } },
See ensureIndex() (page 921) CREATE INDEX idx_user_id_asc db.users.ensureIndex( { user_id: 1 } ) and indexes (page 329) for more inON users(user_id) formation. See ensureIndex() (page 921) CREATE INDEX db.users.ensureIndex( { user_id: 1, age: -1 } ) and indexes (page 329) for more inidx_user_id_asc_age_desc formation. ON users(user_id, age DESC) DROP TABLE users db.users.drop() See drop() (page 921) for more information.
63.4. SQL to MongoDB Mapping Chart
1017
Insert The following table presents the various SQL statements related to inserting records into tables and the corresponding MongoDB statements. SQL INSERT Statements MongoDB insert() Statements Reference See insert() (page 936) for more information.
INSERT INTO users(user_id, db.users.insert( { age, user_id: "bcd001", status) age: 45, VALUES ("bcd001", status: "A" 45, } ) "A")
Select The following table presents the various SQL statements related to reading records from tables and the corresponding MongoDB statements.
1018
SQL SELECT Statements SELECT * FROM users
MongoDB nd() Statements db.users.find()
Reference See find() (page 924) for more information.
See find() (page 924) for more inSELECT id, user_id, status db.users.find( formation. FROM users { }, { user_id: 1, status: 1 } ) SELECT user_id, status FROM users See find() (page 924) for more indb.users.find( formation. { }, { user_id: 1, status: 1, _id: 0 } ) db.users.find( { status: "A" } ) See find() (page 924) for more information.
SELECT * FROM users WHERE status = "A" SELECT user_id, status FROM users WHERE status = "A"
See find() (page 924) for more indb.users.find( formation. { status: "A" }, { user_id: 1, status: 1, _id: 0 } ) See find() (page 924) and $ne db.users.find( (page 767) for more information. { status: { $ne: "A" } } ) db.users.find( { status: "A", age: 50 } ) See find() (page 924) and $and (page 768) for more information.
SELECT * FROM users WHERE status != "A" SELECT * FROM users WHERE status = "A" AND age = 50 SELECT * FROM users WHERE status = "A" OR age = 50 SELECT * FROM users WHERE age > 25 SELECT * FROM users WHERE age < 25 SELECT * FROM users WHERE age > 25 AND age <= 50
See find() (page 924) and $or db.users.find( (page 770) for more information. { $or: [ { status: "A" } , { age: 50 } ] } ) db.users.find( { age: { $gt: 25 } } ) db.users.find( { age: { $lt: 25 } } ) See find() (page 924) and $gt (page 764) for more information.
See find() (page 924) and $lt (page 766) for more information.
See find() (page 924), $gt db.users.find( (page 764), and $lte (page 766) for { age: { $gt: 25, $lte: 50 } } more information. ) 1019 See find() (page 924) and $regex (page 774) for more information.
63.4. SQL to MongoDB Mapping Chart SELECT * db.users.find( FROM users { user_id: /bc/ } WHERE user_id like "%bc%" )
Update Records The following table presents the various SQL statements related to updating existing records in tables and the corresponding MongoDB statements. SQL Update Statements UPDATE users SET status = "C" WHERE age > 25 Reference See update() (page 948), $gt db.users.update( (page 764), and $set (page 788) for { age: { $gt: 25 } }, more information. { $set: { status: "C" } }, { multi: true } ) db.users.update( { status: "A" } , { $inc: { age: 3 } }, { multi: true } ) See update() (page 948), $inc (page 784), and $set (page 788) for more information. MongoDB update() Statements
UPDATE users SET age = age + 3 WHERE status = "A"
Delete Records The following table presents the various SQL statements related to deleting records from tables and the corresponding MongoDB statements. SQL Delete Statements DELETE FROM users WHERE status = "D" DELETE FROM users MongoDB remove() Statements Reference See remove() (page 944) for more db.users.remove( { status: "D" } ) information. See remove() (page 944) for more information.
db.users.remove( )
1020
CHAPTER 64
Architecture and Components
64.1 MongoDB Package Components

64.1.1 Core Processes
The core components in the MongoDB package are: mongod (page 1021), the core database process; mongos (page 1032) the controller and query router for sharded clusters; and mongo (page 1036) the interactive MongoDB Shell. mongod
Synopsis
mongod (page 1021) is the primary daemon process for the MongoDB system. It handles data requests, manages data format, and performs background management operations. This document provides a complete overview of all command line options for mongod (page 1021). These options are primarily useful for testing purposes. In common operation, use the conguration le options (page 1078) to control the behavior of your database, which is fully capable of all operations described below.
Options
mongod --help, -h Returns a basic help and usage text. --version Returns the version of the mongod (page 1021) daemon. --config <filename>, -f <filename> Species a conguration le, that you can use to specify runtime-congurations. While the options are equivalent and accessible via the other command line arguments, the conguration le is the preferred method for runtime conguration of mongod. See the Conguration File Options (page 1078) document for more information about these options. --verbose, -v Increases the amount of internal reporting returned on standard output or in the log le specied by --logpath (page 1022). Use the -v form to control the level of verbosity by including the option multiple times, (e.g. -vvvvv.) 1021
--quiet Runs the mongod (page 1021) instance in a quiet mode that attempts to limit the amount of output. This option suppresses: output from database commands, including drop (page 864), dropIndexes (page 864), diagLogging (page 879), validate (page 907), and clean (page 856). replication activity. connection accepted events. connection closed events. --port <port> Species a TCP port for the mongod (page 1021) to listen for client connections. By default mongod (page 1021) listens for connections on port 27017. UNIX-like systems require root privileges to use ports with numbers lower than 1024. --bind_ip <ip address> The IP address that the mongod (page 1021) process will bind to and listen for connections. By default mongod (page 1021) listens for connections all interfaces. You may attach mongod (page 1021) to any interface; however, when attaching mongod (page 1021) to a publicly accessible interface ensure that you have implemented proper authentication and/or rewall restrictions to protect the integrity of your database. --maxConns <number> Species the maximum number of simultaneous connections that mongod (page 1021) will accept. This setting will have no effect if it is higher than your operating systems congured maximum connection tracking threshold. Note: You cannot set maxConns (page 1079) to a value higher than 20000. --objcheck Forces the mongod (page 1021) to validate all requests from clients upon receipt to ensure that clients never insert invalid documents into the database. For objects with a high degree of sub-document nesting, --objcheck (page 1022) can have a small impact on performance. You can set --noobjcheck (page 1022) to disable object checking at run-time. Changed in version 2.4: MongoDB enables --objcheck (page 1022) by default, to prevent any client from inserting malformed or invalid BSON into a MongoDB database. --noobjcheck New in version 2.4. Disables the default document validation that MongoDB performs on all incoming BSON documents. --logpath <path> Specify a path for the log le that will hold all diagnostic logging information. Unless specied, mongod (page 1021) will output all log information to the standard output. Additionally, unless you also specify --logappend (page 1022), the logle will be overwritten when the process restarts. Note: The behavior of the logging system may change in the near future in response to the SERVER-4499 case. --logappend When specied, this option ensures that mongod (page 1021) appends new entries to the end of the logle rather than overwriting the content of the log when the process restarts. --syslog New in version 2.1.0. 1022 Chapter 64. Architecture and Components
Sends all logging output to the hosts syslog system rather than to standard output or a log le as with --logpath (page 1022). Warning: You cannot use --syslog (page 1022) with --logpath (page 1022). --pidfilepath <path> Specify a le location to hold the PID or process ID of the mongod (page 1021) process. Useful for tracking the mongod (page 1021) process in combination with the mongod --fork (page 1023) option. Without a specied --pidfilepath (page 1023) option, mongos (page 1032) creates no PID le. --keyFile <file> Specify the path to a key le to store authentication information. This option is only useful for the connection between replica set members. See also: Replica Set Security (page 393) and Replica Set Administration (page 419). --nounixsocket Disables listening on the UNIX socket. mongos (page 1032) always listens on the UNIX socket, unless --nounixsocket (page 1023) is set, --bind_ip (page 1022) is not set, or --bind_ip (page 1022) species 127.0.0.1. --unixSocketPrefix <path> Species a path for the UNIX socket. Unless this option has a value mongod (page 1021) creates a socket with http://docs.mongodb.org/manual/tmp as a prex. MongoDB will always create and listen on a UNIX socket, unless --nounixsocket (page 1023) is set, --bind_ip (page 1022) is not set, or --bind_ip (page 1022) species 127.0.0.1. --fork Enables a daemon mode for mongod (page 1021) that runs the process to the background. This is the normal mode of operation, in production and production-like environments, but may not be desirable for testing. --auth Enables database authentication for users connecting from remote hosts. Congure users via the mongo shell (page 1036). If no users exist, the localhost interface will continue to have access to the database until the you create the rst user. See the Security and Authentication (page 131) page for more information regarding this functionality. --cpu Forces mongod (page 1021) to report the percentage of CPU time in write lock. mongod (page 1021) generates output every four seconds. MongoDB writes this data to standard output or the logle if using the logpath (page 1080) option.
--dbpath <path> Specify a directory for the mongod (page 1021) instance to store its data. Typical locations include: http://docs.mongodb.org/manual/srv/mongodb, http://docs.mongodb.org/manual/var/lib/mongodb or http://docs.mongodb.org/manual/opt/mong Unless specied, mongod (page 1021) will look for data les in the default http://docs.mongodb.org/manual/data/db directory. (Windows systems use the \data\db directory.) If you installed using a package management system. Check the http://docs.mongodb.org/manual/etc/mongodb.conf le provided by your packages to see the conguration of the dbpath (page 1081). --diaglog <value> Creates a very verbose, diagnostic log for troubleshooting and recording various errors. MongoDB writes these
64.1. MongoDB Package Components
1023
log les in the dbpath (page 1081) directory in a series of les that begin with the string diaglog and end with the initiation time of the logging as a hex string. The specied value congures the level of verbosity. Possible values, and their impact are as follows. Value 0 1 2 3 7 Setting off. No logging. Log write operations. Log read operations. Log both read and write operations. Log write and some read operations.
You can use the mongosniff (page 1071) tool to replay this output for investigation. Given a typical diaglog le, located at http://docs.mongodb.org/manual/data/db/diaglog.4f76a58c, you might use a command in the following form to read these les:
mongosniff --source DIAGLOG /data/db/diaglog.4f76a58c
--diaglog (page 1023) is for internal use and not intended for most users. Warning: Setting the diagnostic level to 0 will cause mongod (page 1021) to stop writing data to the diagnostic log le. However, the mongod (page 1021) instance will continue to keep the le open, even if it is no longer writing data to the le. If you want to rename, move, or delete the diagnostic log you must cleanly shut down the mongod (page 1021) instance before doing so. --directoryperdb Alters the storage pattern of the data directory to store each databases les in a distinct folder. This option will create directories within the --dbpath (page 1023) named for each directory. Use this option in conjunction with your le system and device conguration so that MongoDB will store data on a number of distinct disk devices to increase write throughput or disk capacity.
1024
Chapter 64. Architecture and Components
Warning: If you have an existing mongod (page 1021) instance and dbpath (page 1081), and you want to enable --directoryperdb (page 1024), you must migrate your existing databases to directories before setting --directoryperdb (page 1024) to access those databases. Example Given a dbpath (page 1081) directory with the following items:
journal mongod.lock local.0 local.1 local.ns test.0 test.1 test.ns
To enable --directoryperdb (page 1024) you would need to modify the dbpath (page 1081) to resemble the following:
journal mongod.lock local/local.0 local/local.1 local/local.ns test/test.0 test/test.1 test/test.ns
--journal Enables operation journaling to ensure write durability and data consistency. mongod (page 1021) enables journaling by default on 64-bit builds of versions after 2.0. --journalOptions <arguments> Provides functionality for testing. Not for general use, and may affect database integrity. --journalCommitInterval <value> Species the maximum amount of time for mongod (page 1021) to allow between journal operations. Possible values are between 2 and 300 milliseconds. Lower values increase the durability of the journal, at the expense of disk performance. The default journal commit interval is 100 milliseconds if a single block device (e.g. physical volume, RAID device, or LVM volume) contains both the journal and the data les. If different block devices provide the journal and data les the default journal commit interval is 30 milliseconds. To force mongod (page 1021) to commit to the journal more frequently, you can specify j:true. When a write operation with j:true is pending, mongod (page 1021) will reduce journalCommitInterval (page 1083) to a third of the set value. --ipv6 Specify this option to enable IPv6 support. This will allow clients to connect to mongod (page 1021) using IPv6 networks. mongod (page 1021) disables IPv6 support by default in mongod (page 1021) and all utilities. --jsonp Permits JSONP access via an HTTP interface. Consider the security implications of allowing this activity before enabling this option.
1025
--noauth Disable authentication. Currently the default. Exists for future compatibility and clarity. --nohttpinterface Disables the HTTP interface. --nojournal Disables the durability journaling. By default, mongod (page 1021) enables journaling in 64-bit versions after v2.0. --noprealloc Disables the preallocation of data les. This will shorten the start up time in some cases, but can cause signicant performance penalties during normal operations. --noscripting Disables the scripting engine. --notablescan Forbids operations that require a table scan. --nssize <value> Species the default size for namespace les (i.e .ns). This option has no impact on the size of existing namespace les. The maximum size is 2047 megabytes. The default value is 16 megabytes; this provides for approximately 24,000 namespaces. Each collection, as well as each index, counts as a namespace. --profile <level> Changes the level of database proling, which inserts information about operation performance into output of mongod (page 1021) or the log le. The following levels are available: Level 0 1 2 Setting Off. No proling. On. Only includes slow operations. On. Includes all operations.
Proling is off by default. Database proling can impact database performance. Enable this option only after careful consideration. --quota Enables a maximum limit for the number data les each database can have. When running with --quota (page 1026), there are a maximum of 8 data les per database. Adjust the quota with the --quotaFiles (page 1026) option. --quotaFiles <number> Modify limit on the number of data les per database. This option requires the --quota (page 1026) setting. The default value for --quotaFiles (page 1026) is 8. --rest Enables the simple REST API. --repair Runs a repair routine on all databases. This is equivalent to shutting down and running the repairDatabase (page 868) database command on all databases. Warning: In general, if you have an intact copy of your data, such as would exist on a very recent backup or an intact member of a replica set, do not use repairDatabase (page 868) or related options like db.repairDatabase() (page 984) in the mongo (page 1036) shell or mongod --repair (page 1026). Restore from an intact copy of your data.
1026
Note: When using journaling, there is almost never any need to run repairDatabase (page 868). In the event of an unclean shutdown, the server will be able restore the data les to a pristine state automatically. Changed in version 2.1.2. If you run the repair option and have data in a journal le, mongod (page 1021) will refuse to start. In these cases you should start mongod (page 1021) without the --repair (page 1026) option to allow mongod (page 1021) to recover data from the journal. This will complete more quickly and will result in a more consistent and complete data set. To continue the repair operation despite the journal les, shut down mongod (page 1021) cleanly and restart with the --repair (page 1026) option. Note: --repair (page 1026) copies data from the source data les into new data les in the repairpath (page 1085), and then replaces the original data les with the repaired data les. If repairpath (page 1085) is on the same device as dbpath (page 1081), you may interrupt a mongod (page 1021) running --repair (page 1026) without affecting the integrity of the data set. --repairpath <path> Species the root directory containing MongoDB data les, to use for the --repair (page 1026) operation. Defaults to a _tmp directory within the dbpath (page 1081). --setParameter <options> New in version 2.4. Species an option to congure on startup. Specify multiple options with multiple --setParameter (page 1027) options. See mongod Parameters (page 1092) for full documentation of these parameters. The setParameter (page 869) database command provides access to many of these parameters. --setParameter (page 1027) supports the following options: enableLocalhostAuthBypass (page 1093) enableTestCommands (page 1093) journalCommitInterval (page 1093) logLevel (page 1094) logUserIds (page 1094) notablescan (page 1094) quiet (page 1095) replApplyBatchSize (page 1094) replIndexPrefetch (page 1094) supportCompatibilityFormPrivilegeDocuments (page 1094) syncdelay (page 1094) textSearchEnabled (page 1095) traceExceptions (page 1095) --slowms <value> Denes the value of slow, for the --profile (page 1026) option. The database logs all slow queries to the log, even when the proler is not turned on. When the database proler is on, mongod (page 1021) the proler writes to the system.profile collection. See the profile (page 888) command for more information on the database proler.
1027
--smallfiles Enables a mode where MongoDB uses a smaller default le size. Specically, --smallfiles (page 1027) reduces the initial size for data les and limits them to 512 megabytes. --smallfiles (page 1027) also reduces the size of each journal les from 1 gigabyte to 128 megabytes. Use --smallfiles (page 1027) if you have a large number of databases that each holds a small quantity of data. --smallfiles (page 1027) can lead your mongod (page 1021) to create a large number of les, which may affect performance for larger databases. --shutdown Used in control scripts, the --shutdown (page 1028) will cleanly and safely terminate the mongod (page 1021) process. When invoking mongod (page 1021) with this option you must set the --dbpath (page 1023) option either directly or by way of the conguration le (page 1078) and the --config (page 1021) option. --shutdown (page 1028) is only available on Linux systems. --syncdelay <value> mongod (page 1021) writes data very quickly to the journal, and lazily to the data les. --syncdelay (page 1028) controls how much time can pass before MongoDB ushes data to the database les via an fsync operation. The default setting is 60 seconds. In almost every situation you should not set this value and use the default setting. The serverStatus (page 889) command reports the background ush threads status via the backgroundFlushing (page 896) eld. syncdelay (page 1086) has no effect on the journal (page 1083) les or journaling (page 71). Warning: If you set --syncdelay (page 1028) to 0, MongoDB will not sync the memory mapped les to disk. Do not set this value on production systems. --sysinfo Returns diagnostic system information and then exits. The information provides the page size, the number of physical pages, and the number of available physical pages. --upgrade Upgrades the on-disk data format of the les specied by the --dbpath (page 1023) to the latest version, if needed. This option only affects the operation of mongod (page 1021) if the data les are in an old format. Note: In most cases you should not set this value, so you can exercise the most control over your upgrade process. See the MongoDB release notes (on the download page) for more information about the upgrade process. --traceExceptions For internal diagnostic use only. Replication Options --replSet <setname> Use this option to congure replication with replica sets. Specify a setname as an argument to this set. All hosts must have the same set name. See also: Replication (page 385), Replica Set Administration (page 419), and Replica Set Conguration (page 463) --oplogSize <value> Species a maximum size in megabytes for the replication operation log (e.g. oplog.) By mongod (page 1021) 1028 Chapter 64. Architecture and Components
creates an oplog based on the maximum amount of space available. For 64-bit systems, the op log is typically 5% of available disk space. Once the mongod (page 1021) has created the oplog for the rst time, changing --oplogSize (page 1028) will not affect the size of the oplog. --fastsync In the context of replica set replication, set this option if you have seeded this member with a snapshot of the dbpath of another member of the set. Otherwise the mongod (page 1021) will attempt to perform an initial sync, as though the member were a new member. Warning: If the data is not perfectly synchronized and mongod (page 1021) starts with fastsync (page 1088), then the secondary or slave will be permanently out of sync with the primary, which may cause signicant consistency problems. --replIndexPrefetch New in version 2.2. You must use --replIndexPrefetch (page 1029) in conjunction with replSet (page 1088). The default value is all and available options are: none all _id_only By default secondary members of a replica set will load all indexes related to an operation into memory before applying operations from the oplog. You can modify this behavior so that the secondaries will only load the _id index. Specify _id_only or none to prevent the mongod (page 1021) from loading any index into memory. Master-Slave Replication These options provide access to conventional master-slave database replication. While this functionality remains accessible in MongoDB, replica sets are the preferred conguration for database replication. --master Congures mongod (page 1021) to run as a replication master. --slave Congures mongod (page 1021) to run as a replication slave. --source <host><:port> For use with the --slave (page 1029) option, the --source option designates the server that this instance will replicate. --only <arg> For use with the --slave (page 1029) option, the --only option species only a single database to replicate. --slavedelay <value> For use with the --slave (page 1029) option, the --slavedelay option congures a delay in seconds, for this slave to wait to apply operations from the master node. --autoresync For use with the --slave (page 1029) option. When set, --autoresync (page 1029) option allows this slave to automatically resync if it is more than 10 seconds behind the master. This setting may be problematic if the --oplogSize (page 1028) species a too small oplog. If the oplog is not large enough to store the difference in changes between the masters current state and the state of the slave, this instance will forcibly resync itself unnecessarily. When you set the autoresync (page 1089) option to false, the slave will not attempt an automatic resync more than once in a ten minute period.
1029
Sharding Cluster Options --configsvr Declares that this mongod (page 1021) instance serves as the cong database of a sharded cluster. When running with this option, clients will not be able to write data to any database other than config and admin. The default port for a mongod (page 1021) with this option is 27019 and the default --dbpath (page 1023) directory is http://docs.mongodb.org/manual/data/configdb, unless specied. Changed in version 2.2: --configsvr (page 1030) also sets --smallfiles (page 1027). Changed in version 2.4: --configsvr (page 1030) creates a local oplog. Do not use --configsvr (page 1030) with --replSet (page 1028) or --shardsvr (page 1030). Cong servers cannot be a shard server or part of a replica set. --shardsvr Congures this mongod (page 1021) instance as a shard in a partitioned cluster. The default port for these instances is 27018. The only effect of --shardsvr (page 1030) is to change the port number. SSL Options See also:
Connect to MongoDB with SSL (page 75) for full documentation of MongoDBs support. --sslOnNormalPorts New in version 2.2. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Enables SSL for mongod (page 1021). With --sslOnNormalPorts (page 1030), a mongod (page 1021) requires SSL encryption for all connections on the default MongoDB port, or the port specied by --port (page 1022). By default, --sslOnNormalPorts (page 1030) is disabled. --sslPEMKeyFile <filename> New in version 2.2. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the .pem le that contains both the SSL certicate and key. Specify the le name of the .pem le using relative or absolute paths When using --sslOnNormalPorts (page 1030), you must specify --sslPEMKeyFile (page 1030). --sslPEMKeyPassword <value> New in version 2.2. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the password to de-crypt the certicate-key le (i.e. --sslPEMKeyFile (page 1030)). Only use --sslPEMKeyPassword (page 1030) if the certicate-key le is encrypted. In all cases, mongod (page 1021) will redact the password from all logging and reporting output. Changed in version 2.4: --sslPEMKeyPassword (page 1030) is only needed when the private key is encrypted. In earlier versions mongod (page 1021) would require --sslPEMKeyPassword (page 1030) when-
1030
ever using --sslOnNormalPorts (page 1030), even when the private key was not encrypted. --sslCAFile <filename> New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the .pem le that contains the root certicate chain from the Certicate Authority. Specify the le name of the .pem le using relative or absolute paths --sslCRLFile <filename> New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the .pem le that contains the Certicate Revocation List. Specify the le name of the .pem le using relative or absolute paths --sslWeakCertificateValidation New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Disables the requirement for SSL certicate validation, that --sslCAFile (page 1031) enables. With --sslWeakCertificateValidation (page 1031), mongod (page 1021) will accept connections if the client does not present a certicate when establishing the connection. If the client presents a certicate and mongod (page 1021) has --sslWeakCertificateValidation (page 1031) enabled, mongod (page 1021) will validate the certicate using the root certicate chain specied by --sslCAFile (page 1031), and reject clients with invalid certicates. Use --sslWeakCertificateValidation (page 1031) if you have a mixed deployment that includes clients that do not or cannot present certicates to mongod (page 1021). --sslFIPSMode New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. When specied, mongod (page 1021) will use the FIPS mode of the installed OpenSSL library. Your system must have a FIPS compliant OpenSSL library to use --sslFIPSMode (page 1031).
Usage
In common usage, the invocation of mongod (page 1021) will resemble the following in the context of an initialization or control script:
1031
See the Conguration File Options (page 1078) for more information on how to congure mongod (page 1021) using the conguration le. mongos
Synopsis
mongos (page 1032) for MongoDB Shard, is a routing service for MongoDB shard congurations that processes queries from the application layer, and determines the location of this data in the sharded cluster, in order to complete these operations. From the perspective of the application, a mongos (page 1032) instance behaves identically to any other MongoDB instance. Note: Changed in version 2.1. Some aggregation operations using the aggregate (page 808) will cause mongos (page 1032) instances to require more CPU resources than in previous versions. This modied performance prole may dictate alternate architecture decisions if you use the aggregation framework extensively in a sharded environment. See also: Sharding (page 483) and Sharded Cluster Overview (page 485).
Options
mongos --help, -h Returns a basic help and usage text. --version Returns the version of the mongod (page 1021) daemon. --config <filename>, -f <filename> Species a conguration le, that you can use to specify runtime-congurations. While the options are equivalent and accessible via the other command line arguments, the conguration le is the preferred method for runtime conguration of mongod. See the Conguration File Options (page 1078) document for more information about these options. Not all conguration options for mongod (page 1021) make sense in the context of mongos (page 1032). --verbose, -v Increases the amount of internal reporting returned on standard output or in the log le specied by --logpath (page 1033). Use the -v form to control the level of verbosity by including the option multiple times, (e.g. -vvvvv.) --quiet Runs the mongos (page 1032) instance in a quiet mode that attempts to limit the amount of output. --port <port> Species a TCP port for the mongos (page 1032) to listen for client connections. By default mongos (page 1032) listens for connections on port 27017. UNIX-like systems require root access to access ports with numbers lower than 1024.
1032
--bind_ip <ip address> The IP address that the mongos (page 1032) process will bind to and listen for connections. By default mongos (page 1032) listens for connections all interfaces. You may attach mongos (page 1032) to any interface; however, when attaching mongos (page 1032) to a publicly accessible interface ensure that you have implemented proper authentication and/or rewall restrictions to protect the integrity of your database. --maxConns <number> Species the maximum number of simultaneous connections that mongos (page 1032) will accept. This setting will have no effect if the value of this setting is higher than your operating systems congured maximum connection tracking threshold. This is particularly useful for mongos (page 1032) if you have a client that creates a number of collections but allows them to timeout rather than close the collections. When you set maxConns (page 1079), ensure the value is slightly higher than the size of the connection pool or the total number of connections to prevent erroneous connection spikes from propagating to the members of a shard cluster. Note: You cannot set maxConns (page 1079) to a value higher than 20000. --objcheck Forces the mongos (page 1032) to validate all requests from clients upon receipt to ensure that invalid objects are never inserted into the database. This option has a performance impact, and is not enabled by default. --logpath <path> Specify a path for the log le that will hold all diagnostic logging information. Unless specied, mongos (page 1032) will output all log information to the standard output. Additionally, unless you also specify --logappend (page 1033), the logle will be overwritten when the process restarts. --logappend Specify to ensure that mongos (page 1032) appends additional logging data to the end of the logle rather than overwriting the content of the log when the process restarts. --setParameter <options> New in version 2.4. Species an option to congure on startup. Specify multiple options with multiple --setParameter (page 1033) options. See mongod Parameters (page 1092) for full documentation of these parameters. The setParameter (page 869) database command provides access to many of these parameters. --setParameter (page 1033) supports the following options: enableLocalhostAuthBypass (page 1093) enableTestCommands (page 1093) logLevel (page 1094) logUserIds (page 1094) notablescan (page 1094) quiet (page 1095) supportCompatibilityFormPrivilegeDocuments (page 1094) syncdelay (page 1094) textSearchEnabled (page 1095) --syslog New in version 2.1.0. Sends all logging output to the hosts syslog system rather than to standard output or a log le as with --logpath (page 1033). 64.1. MongoDB Package Components 1033
Warning: You cannot use --syslog (page 1033) with --logpath (page 1033). --pidfilepath <path> Specify a le location to hold the PID or process ID of the mongos (page 1032) process. Useful for tracking the mongos (page 1032) process in combination with the mongos --fork (page 1034) option. Without a specied --pidfilepath (page 1034) option, mongos (page 1032) creates no PID le. --keyFile <file> Specify the path to a key le to store authentication information. This option is only useful for the connection between mongos (page 1032) instances and components of the sharded cluster. See also: Security Practices for Sharded Clusters (page 492) --nounixsocket Disables listening on the UNIX socket. mongos (page 1032) always listens on the UNIX socket, unless --nounixsocket (page 1034) is set, --bind_ip (page 1032) is not set, or --bind_ip (page 1032) species 127.0.0.1. --unixSocketPrefix <path> Species a path for the UNIX socket. Unless this option has a value mongos (page 1032) creates a socket with http://docs.mongodb.org/manual/tmp as a prex. MongoDB will always create and listen on a UNIX socket, unless --nounixsocket (page 1034) is set, --bind_ip (page 1032) is not set, or --bind_ip (page 1032) species 127.0.0.1. --fork Enables a daemon mode for mongos (page 1032) which forces the process to the background. This is the normal mode of operation, in production and production-like environments, but may not be desirable for testing. --configdb <config1>,<config2><:port>,<config3> Set this option to specify a conguration database (i.e. cong database) for the sharded cluster. You must specify either 1 conguration server or 3 conguration servers, in a comma separated list. Note: mongos (page 1032) instances read from the rst cong server in the list provided. All mongos (page 1032) instances must specify the hosts to the --configdb (page 1034) setting in the same order. If your conguration databases reside in more that one data center, order the hosts in the --configdb (page 1034) argument so that the cong database that is closest to the majority of your mongos (page 1032) instances is rst servers in the list. Warning: Never remove a cong server from the --configdb (page 1034) parameter, even if the cong server or servers are not available, or ofine. --test This option is for internal testing use only, and runs unit tests without starting a mongos (page 1032) instance. --upgrade This option updates the meta data format used by the cong database. --chunkSize <value> The value of the --chunkSize (page 1034) determines the size of each chunk, in megabytes, of data distributed around the sharded cluster. The default value is 64 megabytes, which is the ideal size for chunks in most deployments: larger chunk size can lead to uneven data distribution, smaller chunk size often leads to inefcient movement of chunks between nodes. However, in some circumstances it may be necessary to set a different chunk size.
1034
This option only sets the chunk size when initializing the cluster for the rst time. If you modify the run-time option later, the new value will have no effect. See the Modify Chunk Size (page 523) procedure if you need to change the chunk size on an existing sharded cluster. --ipv6 Enables IPv6 support to allow clients to connect to mongos (page 1032) using IPv6 networks. MongoDB disables IPv6 support by default in mongod (page 1021) and all utilities. --jsonp Permits JSONP access via an HTTP interface. Consider the security implications of allowing this activity before enabling this option. --noscripting Disables the scripting engine. --nohttpinterface New in version 2.1.2. Disables the HTTP interface. --localThreshold New in version 2.2. --localThreshold (page 1035) affects the logic that mongos (page 1032) uses when selecting replica set members to pass read operations to from clients. Specify a value to --localThreshold (page 1035) in milliseconds. The default value is 15, which corresponds to the default value in all of the client drivers (page 555). When mongos (page 1032) receives a request that permits reads to secondary members, the mongos (page 1032) will: nd the member of the set with the lowest ping time. construct a list of replica set members that is within a ping time of 15 milliseconds of the nearest suitable member of the set. If you specify a value for --localThreshold (page 1035), mongos (page 1032) will construct the list of replica members that are within the latency allowed by this value. The mongos (page 1032) will select a member to read from at random from this list. The ping time used for a set member compared by the --localThreshold (page 1035) setting is a moving average of recent ping times, calculated, at most, every 10 seconds. As a result, some queries may reach members above the threshold until the mongos (page 1032) recalculates the average. See the Member Selection (page 406) section of the read preference (page 402) documentation for more information. --noAutoSplit New in version 2.0.7. --noAutoSplit (page 1035) prevents mongos (page 1032) from automatically inserting metadata splits in a sharded collection. If set on all mongos (page 1032), this will prevent MongoDB from creating new chunks as the data in a collection grows. Because any mongos (page 1032) in a cluster can create a split, to totally disable splitting in a cluster you must set --noAutoSplit (page 1035) on all mongos (page 1032). Warning: With --noAutoSplit (page 1035) enabled, the data in your sharded cluster may become imbalanced over time. Enable with caution.
1035
SSL Options
See also:
Connect to MongoDB with SSL (page 75) for full documentation of MongoDBs support. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongos (page 1032) assumes that the database specied to the --db argument holds the users credentials, unless you specify --authenticationDatabase (page 1036). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongos (page 1032) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. mongo
Description
mongo mongo (page 1036) is an interactive JavaScript shell interface to MongoDB, which provides a powerful interface for systems administrators as well as a way for developers to test queries and operations directly with the database. mongo (page 1036) also provides a fully functional JavaScript environment for use with a MongoDB. This document addresses the basic invocation of the mongo (page 1036) shell and an overview of its usage. See also: In addition to this page, consider The mongo Shell (page 581) section of the manual.
Synopsis
mongo [--shell] [--nodb] [--norc] [--quiet] [--port <port>] [--host <host>] [--eval <JavaSc
Interface
Options --shell Enables the shell interface after evaluating a JavaScript le. If you invoke the mongo (page 1036) command and specify a JavaScript le as an argument, or use --eval (page 1037) to specify JavaScript on the command line, the --shell (page 1036) option provides the user with a shell prompt after the le nishes executing. --nodb Prevents the shell from connecting to any database instances. Later, to connect to a database within the shell, see Opening New Connections (page 595). --norc Prevents the shell from sourcing and evaluating ~/.mongorc.js on start up.
1036
--quiet Silences output from the shell during the connection process. --port <port> Species the port where the mongod (page 1021) or mongos (page 1032) instance is listening. Unless specied mongo (page 1036) connects to mongod (page 1021) instances on port 27017, which is the default mongod (page 1021) port. --host <hostname> species the host where the mongod (page 1021) or mongos (page 1032) is running to connect to as <hostname>. By default mongo (page 1036) will attempt to connect to a MongoDB process running on the localhost. --eval <javascript> Evaluates a JavaScript expression specied as an argument to this option. mongo (page 1036) does not load its own environment when evaluating code: as a result many options of the shell environment are not available. --username <username>, -u <username> Species a username to authenticate to the MongoDB instance. Use in conjunction with the --password (page 1037) option to supply a password. If you specify a username and password but the default database or the specied database do not require authentication, mongo (page 1036) will exit with an exception. --password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the --username (page 1037) option to supply a username. If you specify a --username (page 1037) without the --password (page 1037) option, mongo (page 1036) will prompt for a password interactively, if the mongod (page 1021) or mongos (page 1032) requires authentication. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongo (page 1036) assumes that the database name specied in the db address (page 1038) holds the users credentials, unless you specify --authenticationDatabase (page 1037). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongo (page 1036) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. --ssl Enable connection to a mongod (page 1021) or mongos (page 1032) that has SSL encryption. --sslPEMKeyFile <filename> New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the .pem le that contains both the SSL certicate and key. Specify the le name of the .pem le using relative or absolute paths 64.1. MongoDB Package Components 1037
Required when using the --ssl (page 1037) option if the mongod (page 1021) or mongos (page 1032) has sslCAFile (page 1091) enabled without sslWeakCertificateValidation (page 1092). --sslPEMKeyPassword <value> New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the password to decrypt the root certicate chain specied by --sslPEMKeyFile (page 1037). Only required if the certicate-key le is encrypted. --sslCAFile <filename> New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the .pem le that contains the certicate from the Certicate Authority. Specify the le name of the .pem le using relative or absolute paths --help, -h Returns a basic help and usage text. --version Returns the version of the shell. --verbose Increases the verbosity of the output of the shell during the connection process. --ipv6 Enables IPv6 support that allows mongo (page 1036) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongo (page 1036), disable IPv6 support by default. <db address> Species the database address of the database to connect to. For example:
mongo admin
The above command will connect the mongo (page 1036) shell to the admin database on the local machine. You may specify a remote database instance, with the resolvable hostname or IP address. Separate the database name from the hostname using a http://docs.mongodb.org/manual/ character. See the following examples:
mongo mongodb1.example.net mongo mongodb1/admin mongo 10.8.8.10/test
<file.js> Species a JavaScript le to run and then exit. Must be the last option specied. Use the --shell (page 1036) option to return to a shell after the le nishes running. Files ~/.dbshell mongo (page 1036) maintains a history of commands in the .dbshell le.
1038
Note: mongo (page 1036) does not recorded interaction related to authentication in the history le, including authenticate and db.addUser() (page 967). Warning: Versions of Windows mongo.exe earlier than 2.2.0 will save the .dbshell le in the mongo.exe working directory. ~/.mongorc.js mongo (page 1036) will read the .mongorc.js le from the home directory of the user invoking mongo (page 1036). In the le, users can dene variables, customize the mongo (page 1036) shell prompt, or update information that they would like updated every time they launch a shell. If you use the shell to evaluate a JavaScript le or expression either on the command line with --eval (page 1037) or by specifying a .js le to mongo (page 1038), mongo (page 1036) will read the .mongorc.js le after the JavaScript has nished processing. Specify the --norc (page 1036) option to disable reading .mongorc.js. http://docs.mongodb.org/manual/tmp/mongo_edit<time_t>.js Created by mongo (page 1036) when editing a le. If the le exists mongo (page 1036) will append an integer from 1 to 10 to the time value to attempt to create a unique le. %TEMP%mongo_edit<time_t>.js Created by mongo.exe on Windows when editing a le. If the le exists mongo (page 1036) will append an integer from 1 to 10 to the time value to attempt to create a unique le. Environment EDITOR Species the path to an editor to use with the edit shell command. A JavaScript variable EDITOR will override the value of EDITOR (page 1039). HOME Species the path to the home directory where mongo (page 1036) will read the .mongorc.js le and write the .dbshell le. HOMEDRIVE On Windows systems, HOMEDRIVE (page 1039) species the path the directory where mongo (page 1036) will read the .mongorc.js le and write the .dbshell le. HOMEPATH Species the Windows path to the home directory where mongo (page 1036) will read the .mongorc.js le and write the .dbshell le.
Keyboard Shortcuts
The mongo (page 1036) shell supports the following keyboard shortcuts: Keybinding Up arrow Down-arrow Home End Tab
Function Retrieve previous command from history Retrieve next command from history Go to beginning of the line Go to end of the line Autocomplete method/command Continued on next page
MongoDB accommodates multiple keybinding. Since 2.0, mongo (page 1036) includes support for basic emacs keybindings.
1039
Table 64.1 continued from previous page Keybinding Function Left-arrow Go backward one character Right-arrow Go forward one character Ctrl-left-arrow Go backward one word Ctrl-right-arrow Go forward one word Meta-left-arrow Go backward one word Meta-right-arrow Go forward one word Ctrl-A Go to the beginning of the line Ctrl-B Go backward one character Ctrl-C Exit the mongo (page 1036) shell Ctrl-D Delete a char (or exit the mongo (page 1036) shell) Ctrl-E Go to the end of the line Ctrl-F Go forward one character Ctrl-G Abort Ctrl-J Accept/evaluate the line Ctrl-K Kill/erase the line Ctrl-L or type cls Clear the screen Ctrl-M Accept/evaluate the line Ctrl-N Retrieve next command from history Ctrl-P Retrieve previous command from history Ctrl-R Reverse-search command history Ctrl-S Forward-search command history Ctrl-T Transpose characters Ctrl-U Perform Unix line-discard Ctrl-W Perform Unix word-rubout Ctrl-Y Yank Ctrl-Z Suspend (job control works in linux) Ctrl-H Backward-delete a character Ctrl-I Complete, same as Tab Meta-B Go backward one word Meta-C Capitalize word Meta-D Kill word Meta-F Go forward one word Meta-L Change word to lowercase Meta-U Change word to uppercase Meta-Y Yank-pop Meta-Backspace Backward-kill word Meta-< Retrieve the rst command in command history Meta-> Retrieve the last command in command history
Use
Typically users invoke the shell with the mongo (page 1036) command at the system prompt. Consider the following examples for other scenarios. To connect to a database on a remote host using authentication and a non-standard port, use the following form:
mongo --username <user> --password <pass> --hostname <host> --port 28015
Alternatively, consider the following short form:
1040
mongo -u <user> -p <pass> --host <host> --port 28015
Replace <user>, <pass>, and <host> with the appropriate values for your situation and substitute or omit the --port (page 1037) as needed. To execute a JavaScript le without evaluating the ~/.mongorc.js le before starting a shell session, use the following form:
mongo --shell --norc alternate-environment.js
To print return a query as JSON , from the system prompt using the --eval (page 1037) option, use the following form:
mongo --eval db.collection.find().forEach(printjson)
Use single quotes (e.g. ) to enclose the JavaScript, as well as the additional JavaScript required to generate this output.
64.1.2 Windows Services

The mongod.exe (page 1041) and mongos.exe (page 1042) describe the options available for conguring MongoDB when running as a Windows Service. The mongod.exe (page 1041) and mongos.exe (page 1042) binaries provide a superset of the mongod (page 1021) and mongos (page 1032) options. mongod.exe
Synopsis
mongod.exe (page 1041) is the build of the MongoDB daemon (i.e. mongod (page 1021)) for the Windows platform. mongod.exe (page 1041) has all of the features of mongod (page 1021) on Unix-like platforms and is completely compatible with the other builds of mongod (page 1021). In addition, mongod.exe (page 1041) provides several options for interacting with the Windows platform itself. This document only references options that are unique to mongod.exe (page 1041). All mongod (page 1021) options are available. See the mongod (page 1021) and the Conguration File Options (page 1078) documents for more information regarding mongod.exe (page 1041). To install and use mongod.exe (page 1041), read the Install MongoDB on Windows (page 16) document.
Options
mongod.exe --install Installs mongod.exe (page 1041) as a Windows Service and exits. --remove Removes the mongod.exe (page 1041) Windows Service. If mongod.exe (page 1041) is running, this operation will stop and then remove the service. Note: --remove (page 1041) requires the --serviceName (page 1042) if you congured a non-default --serviceName (page 1042) during the --install (page 1041) operation. --reinstall Removes mongod.exe (page 1041) and reinstalls mongod.exe (page 1041) as a Windows Service. 64.1. MongoDB Package Components 1041
--serviceName <name> Default: MongoDB Set the service name of mongod.exe (page 1041) when running as a Windows Service. Use this name with the net start <name> and net stop <name> operations. You must use --serviceName (page 1042) in conjunction with either the --install (page 1041) or --remove (page 1041) install option. --serviceDisplayName <name> Default: Mongo DB Sets the name listed for MongoDB on the Services administrative application. --serviceDescription <description> Default: MongoDB Server Sets the mongod.exe (page 1041) service description. You must use --serviceDescription (page 1042) in conjunction with the --install (page 1041) option. Note: For descriptions that contain spaces, you must enclose the description in quotes. --serviceUser <user> Runs the mongod.exe (page 1041) service in the context of a certain user. This user must have Log on as a service privileges. You must use --serviceUser (page 1042) in conjunction with the --install (page 1041) option. --servicePassword <password> Sets the password for <user> for mongod.exe (page 1041) when running with the --serviceUser (page 1042) option. You must use --servicePassword (page 1042) in conjunction with the --install (page 1041) option. mongos.exe
Synopsis
mongos.exe (page 1042) is the build of the MongoDB Shard (i.e. mongos (page 1032)) for the Windows platform. mongos.exe (page 1042) has all of the features of mongos (page 1032) on Unix-like platforms and is completely compatible with the other builds of mongos (page 1032). In addition, mongos.exe (page 1042) provides several options for interacting with the Windows platform itself. This document only references options that are unique to mongos.exe (page 1042). All mongos (page 1032) options are available. See the mongos (page 1032) and the Conguration File Options (page 1078) documents for more information regarding mongos.exe (page 1042). To install and use mongos.exe (page 1042), read the Install MongoDB on Windows (page 16) document.
Options
mongos.exe --install Installs mongos.exe (page 1042) as a Windows Service and exits.
1042
--remove Removes the mongos.exe (page 1042) Windows Service. If mongos.exe (page 1042) is running, this operation will stop and then remove the service. Note: --remove (page 1042) requires the --serviceName (page 1043) if you congured a non-default --serviceName (page 1043) during the --install (page 1042) operation. --reinstall Removes mongos.exe (page 1042) and reinstalls mongos.exe (page 1042) as a Windows Service. --serviceName <name> Default: MongoS Set the service name of mongos.exe (page 1042) when running as a Windows Service. Use this name with the net start <name> and net stop <name> operations. You must use --serviceName (page 1043) in conjunction with either the --install (page 1042) or --remove (page 1042) install option. --serviceDisplayName <name> Default: Mongo DB Router Sets the name listed for MongoDB on the Services administrative application. --serviceDescription <description> Default: Mongo DB Sharding Router Sets the mongos.exe (page 1042) service description. You must use --serviceDescription (page 1043) in conjunction with the --install (page 1042) option. Note: For descriptions that contain spaces, you must enclose the description in quotes. --serviceUser <user> Runs the mongos.exe (page 1042) service in the context of a certain user. This user must have Log on as a service privileges. You must use --serviceUser (page 1043) in conjunction with the --install (page 1042) option. --servicePassword <password> Sets the password for <user> for mongos.exe (page 1042) when running with the --serviceUser (page 1043) option. You must use --servicePassword (page 1043) in conjunction with the --install (page 1042) option.
64.1.3 Binary Import and Export Tools

mongodump (page 1044) provides a method for creating BSON dump les from the mongod (page 1021) instances, while mongorestore (page 1048) makes it possible to restore these dumps. bsondump (page 1051) converts BSON dump les into JSON . The mongooplog (page 1053) utility provides the ability to stream oplog entries outside of normal replication.
1043
mongodump
Synopsis
mongodump (page 1044) is a utility for creating a binary export of the contents of a database. Consider using this utility as part an effective backup strategy (page 41). Use mongodump (page 1044) in conjunction with mongorestore (page 1048) to restore databases. mongodump (page 1044) can read data from either mongod or mongos (page 1032) instances, in addition to reading directly from MongoDB data les without an active mongod (page 1021). Note: The format of data created by mongodump (page 1044) tool from the 2.2 distribution or later is different and incompatible with earlier versions of mongod (page 1021). See also: mongorestore (page 1048), Create Backup of a Sharded Cluster with Database Dumps (page 56) and Backup Strategies for MongoDB Systems (page 41).
Options
mongodump --help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times, (e.g. -vvvvv.) --version Returns the version of the mongodump (page 1044) utility and exits. --host <hostname><:port> Species a resolvable hostname for the mongod (page 1021) that you wish to use to create the database dump. By default mongodump (page 1044) will attempt to connect to a MongoDB process ruining on the localhost port number 27017. Optionally, specify a port number to connect a MongoDB instance running on a port other than 27017. To connect to a replica set, use the --host (page 1044) argument with a setname, followed by a slash and a comma-separated list of host names and port numbers. The mongodump (page 1044) utility will, given the seed of at least one connected set member, connect to the primary member of that set. This option would resemble:
mongodump --host repl0/mongo0.example.net,mongo0.example.net:27018,mongo1.example.net,mongo2.exa
You can always connect directly to a single MongoDB instance by specifying the host and port number directly. --port <port> Species the port number, if the MongoDB instance is not running on the standard port. (i.e. 27017) You may also specify a port number using the --host (page 1044) option. --ipv6 Enables IPv6 support that allows mongodump (page 1044) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongodump (page 1044), disable IPv6 support by default.
1044
--ssl New in version 2.4: MongoDB added support for SSL connections to mongod (page 1021) instances in mongodump. Note: SSL support in mongodump is not compiled into the default distribution of MongoDB. See Connect to MongoDB with SSL (page 75) for more information on SSL and MongoDB. Additionally, mongodump does not support connections to mongod (page 1021) instances that require client certicate validation. Allows mongodump (page 1044) to connect to mongod (page 1021) instance over an SSL connection. --username <username>, -u <username> Species a username to authenticate to the MongoDB instance, if your database requires authentication. Use in conjunction with the --password (page 1045) option to supply a password. --password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the --username (page 1045) option to supply a username. If you specify a --username (page 1045) without the --password (page 1045) option, mongodump (page 1044) will prompt for a password interactively. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongodump (page 1044) assumes that the database specied to the --db (page 1045) argument holds the users credentials, unless you specify --authenticationDatabase (page 1045). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongodump (page 1044) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. --dbpath <path> Species the directory of the MongoDB data les. If used, the --dbpath (page 1045) option enables mongodump (page 1044) to attach directly to local data les and copy the data without the mongod (page 1021). To run with --dbpath (page 1045), mongodump (page 1044) needs to restrict access to the data directory: as a result, no mongod (page 1021) can access the same path while the process runs. --directoryperdb Use the --directoryperdb (page 1045) in conjunction with the corresponding option to mongod (page 1021). This option allows mongodump (page 1044) to read data les organized with each database located in a distinct directory. This option is only relevant when specifying the --dbpath (page 1045) option. --journal Allows mongodump (page 1044) operations to use the durability journal to ensure that the export is in a consistent state. This option is only relevant when specifying the --dbpath (page 1045) option. --db <db>, -d <db> Use the --db (page 1045) option to specify a database for mongodump (page 1044) to backup. If you do not
1045
specify a DB, mongodump (page 1044) copies all databases in this instance into the dump les. Use this option to backup or copy a smaller subset of your data. --collection <collection>, -c <collection> Use the --collection (page 1046) option to specify a collection for mongodump (page 1044) to backup. If you do not specify a collection, this option copies all collections in the specied database or instance to the dump les. Use this option to backup or copy a smaller subset of your data. --out <path>, -o <path> Species a directory where mongodump (page 1044) will save the output of the database dump. To output the database dump to standard output, specify a - rather than a path. By default, mongodump (page 1044) will save output les in a directory named dump in the current working directory. --query <json>, -q <json> Provides a query to limit (optionally) the documents included in the output of mongodump (page 1044). --oplog Use this option to ensure that mongodump (page 1044) creates a dump of the database that includes an oplog, to create a point-in-time snapshot of the state of a mongod (page 1021) instance. To restore to a specic point-intime backup, use the output created with this option in conjunction with mongorestore --oplogReplay (page 1050). Without --oplog (page 1046), if there are write operations during the dump operation, the dump will not reect a single moment in time. Changes made to the database during the update process can affect the output of the backup. --oplog (page 1046) has no effect when running mongodump (page 1044) against a mongos (page 1032) instance to dump the entire contents of a sharded cluster. However, you can use --oplog (page 1046) to dump individual shards. Note: --oplog (page 1046) only works against nodes that maintain an oplog. This includes all members of a replica set, as well as master nodes in master/slave replication deployments. --repair Use this option to run a repair option in addition to dumping the database. The repair option attempts to repair a database that may be in an inconsistent state as a result of an improper shutdown or mongod (page 1021) crash. --forceTableScan Forces mongodump (page 1044) to scan the data store directly: typically, mongodump (page 1044) saves entries as they appear in the index of the _id eld. Use --forceTableScan (page 1046) to skip the index and scan the data directly. Typically there are two cases where this behavior is preferable to the default: 1.If you have key sizes over 800 bytes that would not be present in the _id index. 2.Your database uses a custom _id eld. When you run with --forceTableScan (page 1046), mongodump (page 1044) does not use $snapshot (page 806). As a result, the dump produced by mongodump (page 1044) can reect the state of the database at many different points in time. Warning: Use --forceTableScan (page 1046) with extreme caution and consideration. Warning: Changed in version 2.2: When used in combination with fsync (page 865) or db.fsyncLock() (page 977), mongod (page 1021) may block some reads, including those from mongodump (page 1044), when queued write operation waits behind the fsync (page 865) lock.
1046
Behavior
When running mongodump (page 1044) against a mongos (page 1032) instance where the sharded cluster consists of replica sets, the read preference of the operation will prefer reads from secondary members of the set.
Usage
See the Use mongodump and mongorestore to Backup and Restore MongoDB Databases (page 43) for a larger overview of mongodump (page 1044) usage. Also see the mongorestore (page 1047) document for an overview of the mongorestore (page 1048), which provides the related inverse functionality. The following command, creates a dump le that contains only the collection named collection in the database named test. In this case the database is running on the local interface on port 27017:
mongodump --collection collection --db test
In the next example, mongodump (page 1044) creates a backup of the database instance stored in the http://docs.mongodb.org/manual/srv/mongodb directory on the local machine. This requires that no mongod (page 1021) instance is using the http://docs.mongodb.org/manual/srv/mongodb directory.
mongodump --dbpath /srv/mongodb
In the nal example, mongodump (page 1044) creates a database dump located at http://docs.mongodb.org/manual/opt/backup/mongodump-2011-10-24, from a database running on port 37017 on the host mongodb1.example.net and authenticating using the username user and the password pass, as follows:
mongodump --host mongodb1.example.net --port 37017 --username user --password pass --out /opt/backup/
mongorestore
Synopsis
The mongorestore (page 1048) program writes data from a binary database dump created by mongodump (page 1044) to a MongoDB instance. mongorestore (page 1048) can create a new database or add data to an existing database. mongorestore (page 1048) can write data to either mongod or mongos (page 1032) instances, in addition to writing directly to MongoDB data les without an active mongod (page 1021). If you restore to an existing database, mongorestore (page 1048) will only insert into the existing database, and does not perform updates of any kind. If existing documents have the same value _id eld in the target database and collection, mongorestore (page 1048) will not overwrite those documents. Remember the following properties of mongorestore (page 1048) behavior: mongorestore (page 1048) recreates indexes recorded by mongodump (page 1044). all operations are inserts, not updates. mongorestore (page 1048) does not wait for a response from a mongod (page 1021) to ensure that the MongoDB process has received or recorded the operation. The mongod (page 1021) will record any errors to its log that occur during a restore operation, but mongorestore (page 1048) will not receive errors.
1047
Note: The format of data created by mongodump (page 1044) tool from the 2.2 distribution or later is different and incompatible with earlier versions of mongod (page 1021).
Options
mongorestore --help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times (e.g. -vvvvv). --version Returns the version of the mongorestore (page 1048) tool. --host <hostname><:port> Species a resolvable hostname for the mongod (page 1021) to which you want to restore the database. By default mongorestore (page 1048) will attempt to connect to a MongoDB process running on the localhost port number 27017. For an example of --host (page 1048), see Restore a Database with mongorestore (page 44). Optionally, specify a port number to connect a MongoDB instance running on a port other than 27017. To connect to a replica set, you can specify the replica set seed name, and a seed list of set members, in the following format:
<replica_set_name>/<hostname1><:port>,<hostname2:<port>,...
--port <port> Species the port number, if the MongoDB instance is not running on the standard port (i.e. 27017). You may also specify a port number using the --host (page 1048) command. For an example of --port (page 1048), see Restore a Database with mongorestore (page 44). --ipv6 Enables IPv6 support that allows mongorestore (page 1048) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongorestore (page 1048), disable IPv6 support by default. --ssl New in version 2.4: MongoDB added support for SSL connections to mongod (page 1021) instances in mongorestore. Note: SSL support in mongorestore is not compiled into the default distribution of MongoDB. See Connect to MongoDB with SSL (page 75) for more information on SSL and MongoDB. Additionally, mongorestore does not support connections to mongod (page 1021) instances that require client certicate validation. Allows mongorestore (page 1048) to connect to mongod (page 1021) instance over an SSL connection. --username <username>, -u <username> Species a username to authenticate to the MongoDB instance, if your database requires authentication. Use in conjunction with the --password (page 1048) option to supply a password. For an example of --username (page 1048), see Restore a Database with mongorestore (page 44).
1048
--password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the --username (page 1048) option to supply a username. For an example of --password (page 1048), see Restore a Database with mongorestore (page 44). If you specify a --username (page 1048) without the --password (page 1048) option, mongorestore (page 1048) will prompt for a password interactively. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongorestore (page 1048) assumes that the database specied to the --db (page 1049) argument holds the users credentials, unless you specify --authenticationDatabase (page 1049). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongorestore (page 1048) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. --dbpath <path> Species the directory of the MongoDB data les. If used, the --dbpath (page 1049) option enables mongorestore (page 1048) to attach directly to local data les and insert the data without the mongod (page 1021). To run with --dbpath (page 1049), mongorestore (page 1048) needs to lock access to the data directory: as a result, no mongod (page 1021) can access the same path while the process runs. For an example of --dbpath (page 1049), see Restore without a Running mongod (page 45). --directoryperdb Use the --directoryperdb (page 1049) in conjunction with the corresponding option to mongod (page 1021), which allows mongorestore (page 1048) to import data into MongoDB instances that have every databases les saved in discrete directories on the disk. This option is only relevant when specifying the --dbpath (page 1049) option. --journal Allows mongorestore (page 1048) write to the durability journal to ensure that the data les will remain in a consistent state during the write process. This option is only relevant when specifying the --dbpath (page 1049) option. For an example of --journal (page 1049), see Restore without a Running mongod (page 45). --db <db>, -d <db> Use the --db (page 1049) option to specify a database for mongorestore (page 1048) to restore data into. If the database doesnt exist, mongorestore (page 1048) will create the specied database. If you do not specify a <db>, mongorestore (page 1048) creates new databases that correspond to the databases where data originated and data may be overwritten. Use this option to restore data into a MongoDB instance that already has data. --db (page 1049) does not control which BSON les mongorestore (page 1048) restores. You must use the mongorestore (page 1048) path option (page 1051) to limit that restored data. --collection <collection>, -c <collection> Use the --collection (page 1049) option to specify a collection for mongorestore (page 1048) to restore. If you do not specify a <collection>, mongorestore (page 1048) imports all collections created.
1049
Existing data may be overwritten. Use this option to restore data into a MongoDB instance that already has data, or to restore only some data in the specied imported data set. --objcheck Forces the mongorestore (page 1048) to validate all requests from clients upon receipt to ensure that clients never insert invalid documents into the database. For objects with a high degree of sub-document nesting, --objcheck (page 1050) can have a small impact on performance. You can set --noobjcheck (page 1050) to disable object checking at run-time. Changed in version 2.4: MongoDB enables --objcheck (page 1050) by default, to prevent any client from inserting malformed or invalid BSON into a MongoDB database. --noobjcheck New in version 2.4. Disables the default document validation that MongoDB performs on all incoming BSON documents. --filter <JSON> Limits the documents that mongorestore (page 1048) imports to only those documents that match the JSON document specied as <JSON>. Be sure to include the document in single quotes to avoid interaction with your systems shell environment. For an example of --filter (page 1050), see Restore a Subset of data from a Binary Database Dump (page 45). --drop Modies the restoration procedure to drop every collection from the target database before restoring the collection from the dumped backup. --oplogReplay Replays the oplog after restoring the dump to ensure that the current state of the database reects the pointin-time backup captured with the mongodump --oplog (page 1046) command. For an example of --oplogReplay (page 1050), see Restore Point in Time Oplog Backup (page 45). --keepIndexVersion Prevents mongorestore (page 1048) from upgrading the index to the latest version during the restoration process. --w <number of replicas per write> New in version 2.2. Species the write concern for each write operation that mongorestore (page 1048) writes to the target database. By default, mongorestore (page 1048) does not wait for a response for write acknowledgment (page 399). --noOptionsRestore New in version 2.2. Prevents mongorestore (page 1048) from setting the collection options, such as those specied by the collMod (page 857) database command, on restored collections. --noIndexRestore New in version 2.2. Prevents mongorestore (page 1048) from restoring and building indexes as specied in the corresponding mongodump (page 1044) output. --oplogLimit <timestamp> New in version 2.2. Prevents mongorestore (page 1048) from applying oplog entries newer than the <timestamp>. Specify <timestamp> values in the form of <time_t>:<ordinal>, where <time_t> is the seconds since the UNIX epoch, and <ordinal> represents a counter of operations in the oplog that occurred in the specied second.
1050
You must use --oplogLimit (page 1050) in conjunction with the --oplogReplay (page 1050) option. <path> The nal argument of the mongorestore (page 1048) command is a directory path. This argument species the location of the database dump from which to restore.
Usage
See Use mongodump and mongorestore to Backup and Restore MongoDB Databases (page 43) for a larger overview of mongorestore (page 1048) usage. Also see the mongodump (page 1044) document for an overview of the mongodump (page 1044), which provides the related inverse functionality. Consider the following example:
mongorestore --collection people --db accounts dump/accounts/people.bson
Here, mongorestore (page 1048) reads the database dump in the dump/ sub-directory of the current directory, and restores only the documents in the collection named people from the database named accounts. mongorestore (page 1048) restores data to the instance running on the localhost interface on port 27017. In the next example, mongorestore (page 1048) restores a backup of the database instance located in dump to a database instance stored in the http://docs.mongodb.org/manual/srv/mongodb on the local machine. This requires that there are no active mongod (page 1021) instances attached to http://docs.mongodb.org/manual/srv/mongodb data directory.
mongorestore --dbpath /srv/mongodb
In the nal example, mongorestore (page 1048) restores a database dump located at http://docs.mongodb.org/manual/opt/backup/mongodump-2011-10-24, from a database running on port 37017 on the host mongodb1.example.net. mongorestore (page 1048) authenticates to the this MongoDB instance using the username user and the password pass, as follows:
mongorestore --host mongodb1.example.net --port 37017 --username user --password pass /opt/backup/mon
bsondump
Synopsis
The bsondump (page 1051) converts BSON les into human-readable formats, including JSON . For example, bsondump (page 1051) is useful for reading the output les generated by mongodump (page 1044). Important: bsondump (page 1051) is a diagnostic tool for inspecting BSON les, not a tool for data ingestion or other application use.
Options
bsondump --help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times, (e.g. -vvvvv.)
1051
--version Returns the version of the bsondump (page 1051) utility. --objcheck Validates each BSON object before outputting it in JSON format. By default, bsondump (page 1051) enables --objcheck (page 1052) by default. For objects with a high degree of sub-document nesting, --objcheck (page 1052) can have a small impact on performance. You can set --noobjcheck (page 1052) to disable object checking. Changed in version 2.4: MongoDB enables --objcheck (page 1052) by default, to prevent any client from inserting malformed or invalid BSON into a MongoDB database. --noobjcheck New in version 2.4. Disables the default document validation that bsondump (page 1051) performs on all BSON documents. --filter <JSON> Limits the documents that bsondump (page 1051) exports to only those documents that match the JSON document specied as <JSON>. Be sure to include the document in single quotes to avoid interaction with your systems shell environment. --type <=json|=debug> Changes the operation of bsondump (page 1051) from outputting JSON (the default) to a debugging format. <bsonfilename> The nal argument to bsondump (page 1051) is a document containing BSON . This data is typically generated by mongodump (page 1044) or by MongoDB in a rollback operation.
Usage
By default, bsondump (page 1051) outputs data to standard output. To create corresponding JSON les, you will need to use the shell redirect. See the following command:
bsondump collection.bson > collection.json
Use the following command (at the system shell) to produce debugging output for a BSON le:
bsondump --type=debug collection.bson
mongooplog New in version 2.2.

Synopsis
mongooplog (page 1053) is a simple tool that polls operations from the replication oplog of a remote server, and applies them to the local server. This capability supports certain classes of real-time migrations that require that the source server remain online and in operation throughout the migration process. Typically this command will take the following form:
mongooplog --from mongodb0.example.net --host mongodb1.example.net
This command copies oplog entries from the mongod (page 1021) instance running on the host mongodb0.example.net and duplicates operations to the host mongodb1.example.net. If you do not need
1052
to keep the --from (page 1054) host running during the migration, consider using mongodump (page 1044) and mongorestore (page 1048) or another backup (page 41) operation, which may be better suited to your operation. Note: If the mongod (page 1021) instance specied by the --from (page 1054) argument is running with authentication (page 1081), then mongooplog (page 1053) will not be able to copy oplog entries. See also: mongodump (page 1044), mongorestore (page 1048), Backup Strategies for MongoDB Systems (page 41), Oplog Internals Overview (page 408), and Replica Set Oplog Sizing (page 392).
Options
mongooplog --help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times, (e.g. -vvvvv.) --version Returns the version of the mongooplog (page 1053) utility. --host <hostname><:port>, -h Species a resolvable hostname for the mongod (page 1021) instance to which mongooplog (page 1053) will apply oplog operations retrieved from the serve specied by the --from (page 1054) option. mongooplog (page 1053) assumes that all target mongod (page 1021) instances are accessible by way of port 27017. You may, optionally, declare an alternate port number as part of the hostname argument. You can always connect directly to a single mongod (page 1021) instance by specifying the host and port number directly. To connect to a replica set, you can specify the replica set seed name, and a seed list of set members, in the following format:
--port Species the port number of the mongod (page 1021) instance where mongooplog (page 1053) will apply oplog entries. Only specify this option if the MongoDB instance that you wish to connect to is not running on the standard port. (i.e. 27017) You may also specify a port number using the --host (page 1053) command. --ipv6 Enables IPv6 support that allows mongooplog (page 1053) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongooplog (page 1053), disable IPv6 support by default. --ssl New in version 2.4: MongoDB added support for SSL connections to mongod (page 1021) instances in mongooplog. Note: SSL support in mongooplog is not compiled into the default distribution of MongoDB. See Connect to MongoDB with SSL (page 75) for more information on SSL and MongoDB. Additionally, mongooplog does not support connections to mongod (page 1021) instances that require client certicate validation.
1053
Allows mongooplog (page 1053) to connect to mongod (page 1021) instance over an SSL connection. --username <username>, -u <username> Species a username to authenticate to the MongoDB instance, if your database requires authentication. Use in conjunction with the --password (page 1054) option to supply a password. --password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the --username (page 1054) option to supply a username. If you specify a --username (page 1054) without the --password (page 1054) option, mongooplog (page 1053) will prompt for a password interactively. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongooplog (page 1053) assumes that the database specied to the --db argument holds the users credentials, unless you specify --authenticationDatabase (page 1054). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongooplog (page 1053) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. --dbpath <path> Species a directory, containing MongoDB data les, to which mongooplog (page 1053) will apply operations from the oplog of the database specied with the --from (page 1054) option. When used, the --dbpath (page 1054) option enables mongo (page 1036) to attach directly to local data les and write data without a running mongod (page 1021) instance. To run with --dbpath (page 1054), mongooplog (page 1053) needs to restrict access to the data directory: as a result, no mongod (page 1021) can be access the same path while the process runs. --directoryperdb Use the --directoryperdb (page 1054) in conjunction with the corresponding option to mongod (page 1021). This option allows mongooplog (page 1053) to write to data les organized with each database located in a distinct directory. This option is only relevant when specifying the --dbpath (page 1054) option. --journal Allows mongooplog (page 1053) operations to use the durability journal to ensure that the data les will remain in a consistent state during the writing process. This option is only relevant when specifying the --dbpath (page 1054) option. --seconds <number>, -s <number> Specify a number of seconds of operations for mongooplog (page 1053) to pull from the remote host (page 1054). Unless specied the default value is 86400 seconds, or 24 hours. --from <host[:port]> Specify the host for mongooplog (page 1053) to retrieve oplog operations from. mongooplog (page 1053) requires this option.
1054
Unless you specify the --host (page 1053) option, mongooplog (page 1053) will apply the operations collected with this option to the oplog of the mongod (page 1021) instance running on the localhost interface connected to port 27017. --oplogns <namespace> Specify a namespace in the --from (page 1054) host where the oplog resides. The default value is local.oplog.rs, which is the where replica set members store their operation log. However, if youve copied oplog entries into another database or collection, use this option to copy oplog entries stored in another location. Namespaces take the form of [database].[collection]. Usage Consider the following prototype mongooplog (page 1053) command:
mongooplog --from mongodb0.example.net --host mongodb1.example.net
Here, entries from the oplog of the mongod (page 1021) running on port 27017. This only pull entries from the last 24 hours. Use the --seconds (page 1054) argument to capture a greater or smaller amount of time. Consider the following example:
mongooplog --from mongodb0.example.net --seconds 172800
In this operation, mongooplog (page 1053) captures 2 full days of operations. To migrate 12 hours of oplog entries, use the following form:
mongooplog --from mongodb0.example.net --seconds 43200
For the previous two examples, mongooplog (page 1053) migrates entries to the mongod (page 1021) process running on the localhost interface connected to the 27017 port. mongooplog (page 1053) can also operate directly on MongoDBs data les if no mongod (page 1021) is running on the target host. Consider the following example:
mongooplog --from mongodb0.example.net --dbpath /srv/mongodb --journal
Here, mongooplog (page 1053) imports oplog operations from the mongod (page 1021) host connected to port 27017. This migrates operations to the MongoDB data les stored in the http://docs.mongodb.org/manual/srv/mongodb directory. Additionally mongooplog (page 1053) will use the durability journal to ensure that the data les remain in a consistent state.
64.1.4 Data Import and Export Tools

mongoimport (page 1056) provides a method for taking data in JSON , CSV , or TSV and importing it into a mongod (page 1021) instance. mongoexport (page 1059) provides a method to export data from a mongod (page 1021) instance into JSON, CSV, or TSV. Note: The conversion between BSON and other formats lacks full type delity. Therefore you cannot use mongoimport (page 1056) and mongoexport (page 1059) for round-trip import and export operations.
mongoimport
Synopsis
The mongoimport (page 1056) tool provides a route to import content from a JSON, CSV, or TSV export created by mongoexport (page 1059), or potentially, another third-party export tool. See the Import and Export MongoDB 64.1. MongoDB Package Components 1055
Data (page 105) document for a more in depth usage overview, and the mongoexport (page 1059) document for more information regarding mongoexport (page 1059), which provides the inverse importing capability. Note: Do not use mongoimport (page 1056) and mongoexport (page 1059) for full instance, production backups because they will not reliably capture data type information. Use mongodump (page 1044) and mongorestore (page 1048) as described in Backup Strategies for MongoDB Systems (page 41) for this kind of functionality.
Options
mongoimport --help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times, (e.g. -vvvvv.) --version Returns the version of the mongoimport (page 1056) program. --host <hostname><:port>, -h Species a resolvable hostname for the mongod (page 1021) to which you want to restore the database. By default mongoimport (page 1056) will attempt to connect to a MongoDB process ruining on the localhost port numbered 27017. Optionally, specify a port number to connect a MongoDB instance running on a port other than 27017. To connect to a replica set, use the --host (page 1056) argument with a setname, followed by a slash and a comma-separated list of host and port names. mongoimport (page 1056) will, given the seed of at least one connected set member, connect to primary node of that set. This option would resemble:
--host repl0/mongo0.example.net,mongo0.example.net:27018,mongo1.example.net,mongo2.example.net
You can always connect directly to a single MongoDB instance by specifying the host and port number directly. --port <port> Species the port number, if the MongoDB instance is not running on the standard port. (i.e. 27017) You may also specify a port number using the mongoimport --host (page 1056) command. --ipv6 Enables IPv6 support that allows mongoimport (page 1056) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongoimport (page 1056), disable IPv6 support by default. --ssl New in version 2.4: MongoDB added support for SSL connections to mongod (page 1021) instances in mongoimport. Note: SSL support in mongoimport is not compiled into the default distribution of MongoDB. See Connect to MongoDB with SSL (page 75) for more information on SSL and MongoDB. Additionally, mongoimport does not support connections to mongod (page 1021) instances that require client certicate validation. Allows mongoimport (page 1056) to connect to mongod (page 1021) instance over an SSL connection.
1056
--username <username>, -u <username> Species a username to authenticate to the MongoDB instance, if your database requires authentication. Use in conjunction with the mongoimport --password (page 1057) option to supply a password. --password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the mongoimport --username (page 1056) option to supply a username. If you specify a --username (page 1056) without the --password (page 1057) option, mongoimport (page 1056) will prompt for a password interactively. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongoimport (page 1056) assumes that the database specied to the --db (page 1057) argument holds the users credentials, unless you specify --authenticationDatabase (page 1057). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongoimport (page 1056) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. --dbpath <path> Species the directory of the MongoDB data les. If used, the --dbpath (page 1057) option enables mongoimport (page 1056) to attach directly to local data les and insert the data without the mongod (page 1021). To run with --dbpath, mongoimport (page 1056) needs to lock access to the data directory: as a result, no mongod (page 1021) can access the same path while the process runs. --directoryperdb Use the --directoryperdb (page 1057) in conjunction with the corresponding option to mongod (page 1021), which allows mongoimport (page 1056) to import data into MongoDB instances that have every databases les saved in discrete directories on the disk. This option is only relevant when specifying the --dbpath (page 1057) option. --journal Allows mongoexport (page 1059) write to the durability journal to ensure that the data les will remain in a consistent state during the write process. This option is only relevant when specifying the --dbpath (page 1057) option. --db <db>, -d <db> Use the --db (page 1057) option to specify a database for mongoimport (page 1056) to restore data. If you do not specify a <db>, mongoimport (page 1056) creates new databases that correspond to the databases where data originated and data may be overwritten. Use this option to restore data into a MongoDB instance that already has data, or to restore only some data in the specied backup. --collection <collection>, -c <collection> Use the --collection (page 1057) option to specify a collection for mongorestore (page 1048) to restore. If you do not specify a <collection>, mongoimport (page 1056) imports all collections created. Existing data may be overwritten. Use this option to restore data into a MongoDB instance that already has data, or to restore only some data in the specied imported data set.
1057
--fields <field1<,filed2>>, -f <field1[,filed2]> Specify a comma separated list of eld names when importing csv or tsv les that do not have eld names in the rst (i.e. header) line of the le. --fieldFile <filename> As an alternative to --fields (page 1057) the --fieldFile (page 1058) option allows you to specify a le (e.g. <file>) to that holds a list of eld names if your csv or tsv le does not include eld names in the rst (i.e. header) line of the le. Place one eld per line. --ignoreBlanks In csv and tsv exports, ignore empty elds. If not specied, mongoimport (page 1056) creates elds without values in imported documents. --type <json|csv|tsv> Declare the type of export format to import. The default format is JSON , but its possible to import csv and tsv les. --file <filename> Specify the location of a le containing the data to import. mongoimport (page 1056) will read data from standard input (e.g. stdin.) if you do not specify a le. --drop Modies the importation procedure so that the target instance drops every collection before restoring the collection from the dumped backup. --headerline If using --type csv (page 1058) or --type tsv (page 1058), use the rst line as eld names. Otherwise, mongoimport (page 1056) will import the rst line as a distinct document. --upsert Modies the import process to update existing objects in the database if they match an imported object, while inserting all other objects. If you do not specify a eld or elds using the --upsertFields (page 1058) mongoimport (page 1056) will upsert on the basis of the _id eld. --upsertFields <field1[,field2]> Species a list of elds for the query portion of the upsert. Use this option if the _id elds in the existing documents dont match the eld in the document, but another eld or eld combination can uniquely identify documents as a basis for performing upsert operations. To ensure adequate performance, indexes should exist for this eld or elds. --stopOnError New in version 2.2. Forces mongoimport (page 1056) to halt the import operation at the rst error rather than continuing the operation despite errors. --jsonArray Changed in version 2.2: The limit on document size increased from 4MB to 16MB. Accept import of data expressed with multiple MongoDB document within a single JSON array. Use in conjunction with mongoexport --jsonArray (page 1061) to import data written as a single JSON array. Limited to imports of 16 MB or smaller.
Usage
In this example, mongoimport (page 1056) imports the csv formatted data in the http://docs.mongodb.org/manual/opt/backups/contacts.csv into the collection contacts in 1058 Chapter 64. Architecture and Components
the users database on the MongoDB instance running on the localhost port numbered 27017.
mongoimport --db users --collection contacts --type csv --file /opt/backups/contacts.csv
In the following example, mongoimport (page 1056) imports the data in the JSON formatted le contacts.json into the collection contacts on the MongoDB instance running on the localhost port number 27017. Journaling is explicitly enabled.
mongoimport --collection contacts --file contacts.json --journal
In the next example, mongoimport (page 1056) takes data passed to it on standard input (i.e. with a | pipe.) and imports it into the collection contacts in the sales database is the MongoDB datales located at http://docs.mongodb.org/manual/srv/mongodb/. if the import process encounters an error, the mongoimport (page 1056) will halt because of the --stopOnError (page 1058) option.
mongoimport --db sales --collection contacts --stopOnError --dbpath /srv/mongodb/
In the nal example, mongoimport (page 1056) imports data from the le http://docs.mongodb.org/manual/opt/backups/mdb1-examplenet.json into the collection contacts within the database marketing on a remote MongoDB database. This mongoimport (page 1056) accesses the mongod (page 1021) instance running on the host mongodb1.example.net over port 37017, which requires the username user and the password pass.
mongoimport --host mongodb1.example.net --port 37017 --username user --password pass --collection con
mongoexport
Synopsis
mongoexport (page 1059) is a utility that produces a JSON or CSV export of data stored in a MongoDB instance. See the Import and Export MongoDB Data (page 105) document for a more in depth usage overview, and the mongoimport (page 1055) document for more information regarding the mongoimport (page 1056) utility, which provides the inverse importing capability. Note: Do not use mongoimport (page 1056) and mongoexport (page 1059) for full-scale backups because they may not reliably capture data type information. Use mongodump (page 1044) and mongorestore (page 1048) as described in Backup Strategies for MongoDB Systems (page 41) for this kind of functionality.
Options
mongoexport --help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times, (e.g. -vvvvv.) --version Returns the version of the mongoexport (page 1059) utility. --host <hostname><:port> Species a resolvable hostname for the mongod (page 1021) from which you want to export data. By default
1059
mongoexport (page 1059) attempts to connect to a MongoDB process ruining on the localhost port number 27017. Optionally, specify a port number to connect a MongoDB instance running on a port other than 27017. To connect to a replica set, you can specify the replica set seed name, and a seed list of set members, in the following format:
--port <port> Species the port number, if the MongoDB instance is not running on the standard port. (i.e. 27017) You may also specify a port number using the mongoexport --host (page 1059) command. --ipv6 Enables IPv6 support that allows mongoexport (page 1059) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongoexport (page 1059), disable IPv6 support by default. --ssl New in version 2.4: MongoDB added support for SSL connections to mongod (page 1021) instances in mongoexport. Note: SSL support in mongoexport is not compiled into the default distribution of MongoDB. See Connect to MongoDB with SSL (page 75) for more information on SSL and MongoDB. Additionally, mongoexport does not support connections to mongod (page 1021) instances that require client certicate validation. Allows mongoexport (page 1059) to connect to mongod (page 1021) instance over an SSL connection. --username <username>, -u <username> Species a username to authenticate to the MongoDB instance, if your database requires authentication. Use in conjunction with the mongoexport --password (page 1060) option to supply a password. --password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the --username (page 1060) option to supply a username. If you specify a --username (page 1060) without the --password (page 1060) option, mongoexport (page 1059) will prompt for a password interactively. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongoexport (page 1059) assumes that the database specied to the --db (page 1061) argument holds the users credentials, unless you specify --authenticationDatabase (page 1060). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongoexport (page 1059) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. 1060 Chapter 64. Architecture and Components
--dbpath <path> Species the directory of the MongoDB data les. If used, the --dbpath option enables mongoexport (page 1059) to attach directly to local data les and insert the data without the mongod (page 1021). To run with --dbpath, mongoexport (page 1059) needs to lock access to the data directory: as a result, no mongod (page 1021) can access the same path while the process runs. --directoryperdb Use the --directoryperdb (page 1061) in conjunction with the corresponding option to mongod (page 1021), which allows mongoexport (page 1059) to export data into MongoDB instances that have every databases les saved in discrete directories on the disk. This option is only relevant when specifying the --dbpath (page 1060) option. --journal Allows mongoexport (page 1059) operations to access the durability journal to ensure that the export is in a consistent state. This option is only relevant when specifying the --dbpath (page 1060) option. --db <db>, -d <db> Use the --db (page 1061) option to specify the name of the database that contains the collection you want to export. --collection <collection>, -c <collection> Use the --collection (page 1061) option to specify the collection that you want mongoexport (page 1059) to export. --fields <field1[,field2]>, -f <field1[,field2]> Specify a eld or elds to include in the export. Use a comma separated list of elds to specify multiple elds. For --csv (page 1061) output formats, mongoexport (page 1059) includes only the specied eld(s), and the specied eld(s) can be a eld within a sub-document. For JSON output formats, mongoexport (page 1059) includes only the specied eld(s) and the _id eld, and if the specied eld(s) is a eld within a sub-document, the mongoexport (page 1059) includes the sub-document with all its elds, not just the specied eld within the document. --fieldFile <file> As an alternative to --fields (page 1061), the --fieldFile (page 1061) option allows you to specify in a le the eld or elds to include in the export and is only valid with the --csv (page 1061) option. The le must have only one eld per line, and the line(s) must end with the LF character (0x0A). mongoexport (page 1059) includes only the specied eld(s). The specied eld(s) can be a eld within a sub-document. --query <JSON> Provides a JSON document as a query that optionally limits the documents returned in the export. --csv Changes the export format to a comma separated values (CSV) format. By default mongoexport (page 1059) writes data using one JSON document for every MongoDB document. If you specify --csv (page 1061), then you must also use either the --fields (page 1061) or the --fieldFile (page 1061) option to declare the elds to export from the collection. --jsonArray Modies the output of mongoexport (page 1059) to write the entire contents of the export as a single JSON array. By default mongoexport (page 1059) writes data using one JSON document for every MongoDB document. --slaveOk, -k Allows mongoexport (page 1059) to read data from secondary or slave nodes when using mongoexport (page 1059) with a replica set. This option is only available if connected to a mongod (page 1021) or mongos (page 1032) and is not available when used with the mongoexport --dbpath (page 1060) option.
1061
This is the default behavior. --out <file>, -o <file> Specify a le to write the export to. If you do not specify a le name, the mongoexport (page 1059) writes data to standard output (e.g. stdout). --forceTableScan New in version 2.2. Forces mongoexport (page 1059) to scan the data store directly: typically, mongoexport (page 1059) saves entries as they appear in the index of the _id eld. Use --forceTableScan (page 1062) to skip the index and scan the data directly. Typically there are two cases where this behavior is preferable to the default: 1.If you have key sizes over 800 bytes that would not be present in the _id index. 2.Your database uses a custom _id eld. When you run with --forceTableScan (page 1062), mongoexport (page 1059) does not use $snapshot (page 806). As a result, the export produced by mongoexport (page 1059) can reect the state of the database at many different points in time. Warning: Use --forceTableScan (page 1062) with extreme caution and consideration.
Usage
In the following example, mongoexport (page 1059) exports the collection contacts from the users database from the mongod (page 1021) instance running on the localhost port number 27017. This command writes the export data in CSV format into a le located at http://docs.mongodb.org/manual/opt/backups/contacts.csv. The fields.txt le contains a line-separated list of elds to export.
mongoexport --db users --collection contacts --csv --fieldFile fields.txt --out /opt/backups/contacts
The next example creates an export of the collection contacts from the MongoDB instance running on the localhost port number 27017, with journaling explicitly enabled. This writes the export to the contacts.json le in JSON format.
mongoexport --db sales --collection contacts --out contacts.json --journal
The following example exports the collection contacts from the sales database located in the MongoDB data les located at http://docs.mongodb.org/manual/srv/mongodb/. This operation writes the export to standard output in JSON format.
mongoexport --db sales --collection contacts --dbpath /srv/mongodb/
Warning: The above example will only succeed if there is no mongod (page 1021) connected to the data les located in the http://docs.mongodb.org/manual/srv/mongodb/ directory. The nal example exports the collection contacts from the database marketing . This data resides on the MongoDB instance located on the host mongodb1.example.net running on port 37017, which requires the username user and the password pass.
mongoexport --host mongodb1.example.net --port 37017 --username user --password pass --collection con
1062
64.1.5 Diagnostic Tools

mongostat (page 1063), mongotop (page 1068), and mongosniff (page 1071) provide diagnostic information related to the current operation of a mongod (page 1021) instance. Note: Because mongosniff (page 1071) depends on libpcap, most distributions of MongoDB do not include mongosniff (page 1071).
mongostat
Synopsis
The mongostat (page 1063) utility provides a quick overview of the status of a currently running mongod (page 1021) or mongos (page 1032) instance. mongostat (page 1063) is functionally similar to the UNIX/Linux le system utility vmstat, but provides data regarding mongod (page 1021) and mongos (page 1032) instances. See also: For more information about monitoring MongoDB, see Monitoring for MongoDB (page 91). For more background on various other MongoDB status outputs see: serverStatus (page 889) replSetGetStatus (page 840) dbStats (page 877) collStats (page 873) For an additional utility that provides MongoDB metrics see mongotop (page 1067). mongostat (page 1063) connects to the mongod (page 1021) instance running on the local host interface on TCP port 27017; however, mongostat (page 1063) can connect to any accessible remote mongod (page 1021) instance.
Options
mongostat --help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times, (e.g. -vvvvv.) --version Returns the version of the mongostat (page 1063) utility. --host <hostname><:port> Species a resolvable hostname for the mongod (page 1021) from which you want to export data. By default mongostat (page 1063) attempts to connect to a MongoDB instance running on the localhost port number 27017. Optionally, specify a port number to connect a MongoDB instance running on a port other than 27017. To connect to a replica set, you can specify the replica set seed name, and a seed list of set members, in the following format:
1063
--port <port> Species the port number, if the MongoDB instance is not running on the standard port. (i.e. 27017) You may also specify a port number using the mongostat --host (page 1063) command. --ipv6 Enables IPv6 support that allows mongostat (page 1063) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongostat (page 1063), disable IPv6 support by default. --ssl New in version 2.4: MongoDB added support for SSL connections to mongod (page 1021) instances in mongostat. Note: SSL support in mongostat is not compiled into the default distribution of MongoDB. See Connect to MongoDB with SSL (page 75) for more information on SSL and MongoDB. Additionally, mongostat does not support connections to mongod (page 1021) instances that require client certicate validation. Allows mongostat (page 1063) to connect to mongod (page 1021) instance over an SSL connection. --username <username>, -u <username> Species a username to authenticate to the MongoDB instance, if your database requires authentication. Use in conjunction with the mongostat --password (page 1064) option to supply a password. Important: This user must have sufcient credentials to run the serverStatus (page 889) command, which is the clusterAdmin (page 157) role. See User Privilege Roles in MongoDB (page 155) and system.users Privilege Documents (page 159) for more information. --password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the mongostat --username (page 1064) option to supply a username. If you specify a --username (page 1064) without the --password (page 1064) option, mongostat (page 1063) will prompt for a password interactively. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongostat (page 1063) assumes that the database specied to the --db argument holds the users credentials, unless you specify --authenticationDatabase (page 1064). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongostat (page 1063) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication.
1064
--noheaders Disables the output of column or eld names. --rowcount <number>, -n <number> Controls the number of rows to output. Use in conjunction with the sleeptime argument to control the duration of a mongostat (page 1063) operation. Unless --rowcount (page 1065) is specied, mongostat (page 1063) will return an innite number of rows (e.g. value of 0.) --http Congures mongostat (page 1063) to collect data using the HTTP interface rather than a raw database connection. --discover With this option mongostat (page 1063) discovers and reports on statistics from all members of a replica set or sharded cluster. When connected to any member of a replica set, --discover (page 1065) all nonhidden members of the replica set. When connected to a mongos (page 1032), mongostat (page 1063) will return data from all shards in the cluster. If a replica set provides a shard in the sharded cluster, mongostat (page 1063) will report on non-hidden members of that replica set. The mongostat --host (page 1063) option is not required but potentially useful in this case. --all Congures mongostat (page 1063) to return all optional elds (page 1065). <sleeptime> The nal argument is the length of time, in seconds, that mongostat (page 1063) waits in between calls. By default mongostat (page 1063) returns one call every second. mongostat (page 1063) returns values that reect the operations over a 1 second period. For values of <sleeptime> greater than 1, mongostat (page 1063) averages data to reect average operations per second.
Fields
mongostat (page 1063) returns values that reect the operations over a 1 second period. When mongostat <sleeptime> has a value greater than 1, mongostat (page 1063) averages the statistics to reect average operations per second. mongostat (page 1063) outputs the following elds: inserts The number of objects inserted into the database per second. If followed by an asterisk (e.g. *), the datum refers to a replicated operation. query The number of query operations per second. update The number of update operations per second. delete The number of delete operations per second. getmore The number of get more (i.e. cursor batch) operations per second. command The number of commands per second. On slave and secondary systems, mongostat (page 1063) presents two values separated by a pipe character (e.g. |), in the form of local|replicated commands. 64.1. MongoDB Package Components 1065
flushes The number of fsync operations per second. mapped The total amount of data mapped in megabytes. This is the total data size at the time of the last mongostat (page 1063) call. size The amount of (virtual) memory in megabytes used by the process at the time of the last mongostat (page 1063) call. res The amount of (resident) memory in megabytes used by the process at the time of the last mongostat (page 1063) call. faults Changed in version 2.1. The number of page faults per second. Before version 2.1 this value was only provided for MongoDB instances running on Linux hosts. locked The percent of time in a global write lock. Changed in version 2.2: The locked db eld replaces the locked % eld to more appropriate data regarding the database specic locks in version 2.2. locked db New in version 2.2. The percent of time in the per-database context-specic lock. mongostat (page 1063) will report the database that has spent the most time since the last mongostat (page 1063) call with a write lock. This value represents the amount of time that the listed database spent in a locked state combined with the time that the mongod (page 1021) spent in the global lock. Because of this, and the sampling method, you may see some values greater than 100%. idx miss The percent of index access attempts that required a page fault to load a btree node. This is a sampled value. qr The length of the queue of clients waiting to read data from the MongoDB instance. qw The length of the queue of clients waiting to write data from the MongoDB instance. ar The number of active clients performing read operations. aw The number of active clients performing write operations. netIn The amount of network trafc, in bytes, received by the MongoDB instance. This includes trafc from mongostat (page 1063) itself. netOut The amount of network trafc, in bytes, sent by the MongoDB instance. This includes trafc from mongostat (page 1063) itself.
1066
conn The total number of open connections. set The name, if applicable, of the replica set. repl The replication status of the node. Value M SEC REC UNK SLV
Usage
Replication Type master secondary recovering unknown slave
In the rst example, mongostat (page 1063) will return data every second for 20 seconds. mongostat (page 1063) collects data from the mongod (page 1021) instance running on the localhost interface on port 27017. All of the following invocations produce identical behavior:
mongostat mongostat mongostat mongostat --rowcount 20 1 --rowcount 20 -n 20 1 -n 20
In the next example, mongostat (page 1063) returns data every 5 minutes (or 300 seconds) for as long as the program runs. mongostat (page 1063) collects data from the mongod (page 1021) instance running on the localhost interface on port 27017. Both of the following invocations produce identical behavior.
mongostat --rowcount 0 300 mongostat -n 0 300 mongostat 300
In the following example, mongostat (page 1063) returns data every 5 minutes for an hour (12 times.) mongostat (page 1063) collects data from the mongod (page 1021) instance running on the localhost interface on port 27017. Both of the following invocations produce identical behavior.
mongostat --rowcount 12 300 mongostat -n 12 300
In many cases, using the --discover (page 1065) will help provide a more complete snapshot of the state of an entire group of machines. If a mongos (page 1032) process connected to a sharded cluster is running on port 27017 of the local machine, you can use the following form to return statistics from all members of the cluster:
mongostat --discover
mongotop
Synopsis
mongotop (page 1068) provides a method to track the amount of time a MongoDB instance spends reading and writing data. mongotop (page 1068) provides statistics on a per-collection level. By default, mongotop (page 1068) returns values every second.
1067
See also: For more information about monitoring MongoDB, see Monitoring for MongoDB (page 91). For additional background on various other MongoDB status outputs see: serverStatus (page 889) replSetGetStatus (page 840) dbStats (page 877) collStats (page 873) For an additional utility that provides MongoDB metrics see mongostat (page 1063).
Options
mongotop --help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times, (e.g. -vvvvv.) --version Print the version of the mongotop (page 1068) utility and exit. --host <hostname><:port> Species a resolvable hostname for the mongod from which you want to export data. By default mongotop (page 1068) attempts to connect to a MongoDB process running on the localhost port number 27017. Optionally, specify a port number to connect a MongoDB instance running on a port other than 27017. To connect to a replica set, you can specify the replica set seed name, and a seed list of set members, in the following format:
--port <port> Species the port number, if the MongoDB instance is not running on the standard port. (i.e. 27017) You may also specify a port number using the mongotop --host (page 1068) command. --ipv6 Enables IPv6 support that allows mongotop (page 1068) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongotop (page 1068), disable IPv6 support by default. --username <username>, -u <username> Species a username to authenticate to the MongoDB instance, if your database requires authentication. Use in conjunction with the mongotop (page 1068) option to supply a password. --password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the --username (page 1068) option to supply a username. If you specify a --username (page 1068) without the --password (page 1068) option, mongotop (page 1068) will prompt for a password interactively.
1068
--authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongotop (page 1068) assumes that the database specied to the --db argument holds the users credentials, unless you specify --authenticationDatabase (page 1068). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB. --authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongotop (page 1068) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. --locks New in version 2.2. Toggles the mode of mongotop (page 1068) to report on use of per-database locks (page 890). These data are useful for measuring concurrent operations and lock percentage. <sleeptime> The nal argument is the length of time, in seconds, that mongotop (page 1068) waits in between calls. By default mongotop (page 1068) returns data every second.
Fields
mongotop (page 1068) returns time values specied in milliseconds (ms.) mongotop (page 1068) only reports active namespaces or databases, depending on the --locks (page 1069) option. If you dont see a database or collection, it has received no recent activity. You can issue a simple operation in the mongo (page 1036) shell to generate activity to affect the output of mongotop (page 1068). mongotop.ns Contains the database namespace, which combines the database name and collection. Changed in version 2.2: If you use the --locks (page 1069), the ns (page 1069) eld does not appear in the mongotop (page 1068) output. mongotop.db New in version 2.2. Contains the name of the database. The database named . refers to the global lock, rather than a specic database. This eld does not appear unless you have invoked mongotop (page 1068) with the --locks (page 1069) option. mongotop.total Provides the total amount of time that this mongod (page 1021) spent operating on this namespace. mongotop.read Provides the amount of time that this mongod (page 1021) spent performing read operations on this namespace. mongotop.write Provides the amount of time that this mongod (page 1021) spent performing write operations on this namespace.
1069
mongotop.<timestamp> Provides a time stamp for the returned data.

Use
By default mongotop (page 1068) connects to the MongoDB instance running on the localhost port 27017. However, mongotop (page 1068) can optionally connect to remote mongod (page 1021) instances. See the mongotop options (page 1068) for more information. To force mongotop (page 1068) to return less frequently specify a number, in seconds at the end of the command. In this example, mongotop (page 1068) will return every 15 seconds.
mongotop 15
This command produces the following output:

connected to: 127.0.0.1 ns test.system.namespaces local.system.replset local.system.indexes admin.system.indexes admin. ns test.system.namespaces local.system.replset local.system.indexes admin.system.indexes admin. total 0ms 0ms 0ms 0ms 0ms total 0ms 0ms 0ms 0ms 0ms read 0ms 0ms 0ms 0ms 0ms read 0ms 0ms 0ms 0ms 0ms write 0ms 0ms 0ms 0ms 0ms write 0ms 0ms 0ms 0ms 0ms 2012-08-13T15:45:40
2012-08-13T15:45:55
To return a mongotop (page 1068) report every 5 minutes, use the following command:
mongotop 300
To report the use of per-database locks, use mongotop --locks (page 1069), which produces the following output:
$ mongotop --locks connected to: 127.0.0.1 db local admin . total 0ms 0ms 0ms read 0ms 0ms 0ms write 0ms 0ms 0ms 2012-08-13T16:33:34
mongosniff
Synopsis
mongosniff (page 1071) provides a low-level operation tracing/snifng view into database activity in real time. Think of mongosniff (page 1071) as a MongoDB-specic analogue of tcpdump for TCP/IP network trafc. Typically, mongosniff (page 1071) is most frequently used in driver development. Note: mongosniff (page 1071) requires libpcap and is only available for Unix-like systems. Furthermore, the version distributed with the MongoDB binaries is dynamically linked against aversion 0.9 of libpcap. If your
1070
system has a different version of libpcap, you will need to compile mongosniff (page 1071) yourself or create a symbolic link pointing to libpcap.so.0.9 to your local version of libpcap. Use an operation that resembles the following:
ln -s /usr/lib/libpcap.so.1.1.1 /usr/lib/libpcap.so.0.9
Change the paths and name of the shared library as needed. As an alternative to mongosniff (page 1071), Wireshark, a popular network snifng tool is capable of inspecting and parsing the MongoDB wire protocol.
Options
mongosniff --help Returns a basic help and usage text. --forward <host><:port> Declares a host to forward all parsed requests that the mongosniff (page 1071) intercepts to another mongod (page 1021) instance and issue those operations on that database instance. Specify the target host name and port in the <host><:port> format. To connect to a replica set, you can specify the replica set seed name, and a seed list of set members, in the following format:
--source <NET [interface]>, <FILE [filename]>, <DIAGLOG [filename]> Species source material to inspect. Use --source NET [interface] to inspect trafc from a network interface (e.g. eth0 or lo.) Use --source FILE [filename] to read captured packets in pcap format. You may use the --source DIAGLOG [filename] option to read the output les produced by the --diaglog (page 1023) option. --objcheck Modies the behavior to only display invalid BSON objects and nothing else. Use this option for troubleshooting driver development. This option has some performance impact on the performance of mongosniff (page 1071). <port> Species alternate ports to sniff for trafc. By default, mongosniff (page 1071) watches for MongoDB trafc on port 27017. Append multiple port numbers to the end of mongosniff (page 1071) to monitor trafc on multiple ports.
Usage
Use the following command to connect to a mongod (page 1021) or mongos (page 1032) running on port 27017 and 27018 on the localhost interface:
mongosniff --source NET lo 27017 27018
Use the following command to only log invalid BSON objects for the mongod (page 1021) or mongos (page 1032) running on the localhost interface and port 27018, for driver development and troubleshooting:
1071
mongosniff --objcheck --source NET lo 27018
Build mongosniff
To build mongosniff yourself, Linux users can use the following procedure: 1. Obtain prerequisites using your operating systems package management software. Dependencies include: libpcap - to capture network packets. git - to download the MongoDB source code. scons and a C++ compiler - to build mongosniff (page 1071). 2. Download a copy of the MongoDB source code using git:
git clone git://github.com/mongodb/mongo.git
3. Issue the following sequence of commands to change to the mongo/ directory and build mongosniff (page 1071):
cd mongo scons mongosniff
Note: If you run scons mongosniff before installing libpcap you must run scons clean before you can build mongosniff (page 1071).
mongoperf
Synopsis
mongoperf (page 1072) is a utility to check disk I/O performance independently of MongoDB. It times tests of random disk I/O and presents the results. You can use mongoperf (page 1072) for any case apart from MongoDB. The mmf (page 1073) true mode is completely generic. In that mode is it somewhat analogous to tools such as bonnie++ (albeit mongoperf is simpler). Specify options to mongoperf (page 1072) using a JavaScript document. See also: bonnie bonnie++ Output from an example run Checking Disk Performance with the mongoperf Utility
Options
mongoperf --help Displays the options to mongoperf (page 1072). Specify options to mongoperf (page 1072) with a JSON document described in the Conguration Fields (page 1073) section.
1072
<jsonconfig> mongoperf (page 1072) accepts conguration options in the form of a le that holds a JSON document. You must stream the content of this le into mongoperf (page 1072), as in the following operation:
mongoperf < config
In this example config is the name of a le that holds a JSON document that resembles the following example:
{ nThreads:<n>, fileSizeMB:<n>, sleepMicros:<n>, mmf:<bool>, r:<bool>, w:<bool>, recSizeKB:<n>, syncDelay:<n> }
See the Conguration Fields (page 1073) section for documentation of each of these elds.
Conguration Fields
mongoperf.nThreads Type: Integer. Default: 1 Denes the number of threads mongoperf (page 1072) will use in the test. To saturate your systems storage system you will need multiple threads. Consider setting nThreads (page 1073) to 16. mongoperf.fileSizeMB Type: Integer. Default: 1 megabyte (i.e. 10242 bytes) Test le size. mongoperf.sleepMicros Type: Integer. Default: 0 mongoperf (page 1072) will pause for the number of specied sleepMicros (page 1073) divided by the nThreads (page 1073) between each operation. mongoperf.mmf Type: Boolean. Default: false Set mmf (page 1073) to true to use memory mapped les for the tests. Generally: when mmf (page 1073) is false, mongoperf (page 1072) tests direct, physical, I/O, without caching. Use a large le size to test heavy random I/O load and to avoid I/O coalescing. when mmf (page 1073) is true, mongoperf (page 1072) runs tests of the caching system, and can use normal le system cache. Use mmf (page 1073) in this mode to test le system cache behavior with memory mapped les.
1073
mongoperf.r Type: Boolean. Default: false Set r (page 1073) to true to perform reads as part of the tests. Either r (page 1073) or w (page 1074) must be true. mongoperf.w Type: Boolean. Default: false Set w (page 1074) to true to perform writes as part of the tests. Either r (page 1073) or w (page 1074) must be true. mongoperf.recSizeKB New in version 2.4. Type: Integer. Default: 4 kb The size of each write operation. mongoperf.syncDelay Type: Integer. Default: 0 Seconds between disk ushes. mongoperf.syncDelay (page 1074) is similar to --syncdelay (page 1028) for mongod (page 1021). The syncDelay (page 1074) controls how frequently mongoperf (page 1072) performs an asynchronous disk ush the memory mapped le used for testing. By default, mongod (page 1021) performs this operation every every 60 seconds. Use syncDelay (page 1074) to test basic system performance of this type of operation. Only use syncDelay (page 1074) in conjunction with mmf (page 1073) set to true. The default value of 0 disables this
Use mongoperf < jsonconfigfile
Replace jsonconfigfile with the path to the mongoperf (page 1072) conguration. You may also invoke mongoperf (page 1072) in the following form:
echo "{nThreads:16,fileSizeMB:1000,r:true}" | ./mongoperf
In this operation: mongoperf (page 1072) tests direct physical random read ios, using 16 concurrent reader threads. mongoperf (page 1072) uses a 1 gigabyte test le. Consider using iostat, as invoked in the following example to monitor I/O performance during the test.
iostat -xm 2
1074
64.1.6 GridFS
mongofiles (page 1075) provides a command-line interact to a MongoDB GridFS storage system. mongofiles
Synopsis
The mongofiles (page 1075) utility makes it possible to manipulate les stored in your MongoDB instance in GridFS objects from the command line. It is particularly useful as it provides an interface between objects stored in your le system and GridFS. All mongofiles (page 1075) commands take arguments in three groups: 1. Options (page 1076). You may use one or more of these options to control the behavior of mongofiles (page 1075). 2. Commands (page 1075). Use one of these commands to determine the action of mongofiles (page 1075). 3. A le name representing either the name of a le on your systems le system, a GridFS object. mongofiles (page 1075), like mongodump (page 1044), mongoexport (page 1059), mongoimport (page 1056), and mongorestore (page 1048), can access data stored in a MongoDB data directory without requiring a running mongod (page 1021) instance, if no other mongod (page 1021) is running. Note: For replica sets, mongofiles (page 1075) can only read from the sets primary.
Commands
mongofiles list <prefix> Lists the les in the GridFS store. The characters specied after list (e.g. <prefix>) optionally limit the list of returned items to les that begin with that string of characters. search <string> Lists the les in the GridFS store with names that match any portion of <string>. put <filename> Copy the specied le from the local le system into GridFS storage. Here, <filename> refers to the name the object will have in GridFS, and mongofiles (page 1075) assumes that this reects the name the le has on the local le system. If the local lename is different use the mongofiles --local (page 1077) option. get <filename> Copy the specied le from GridFS storage to the local le system. Here, <filename> refers to the name the object will have in GridFS, and mongofiles (page 1075) assumes that this reects the name the le has on the local le system. If the local lename is different use the mongofiles --local (page 1077) option. delete <filename> Delete the specied le from GridFS storage.
1075
Options
--help Returns a basic help and usage text. --verbose, -v Increases the amount of internal reporting returned on the command line. Increase the verbosity with the -v form by including the option multiple times, (e.g. -vvvvv.) --version Returns the version of the mongofiles (page 1075) utility. --host <hostname><:port> Species a resolvable hostname for the mongod (page 1021) that holds your GridFS system. By default mongofiles (page 1075) attempts to connect to a MongoDB process ruining on the localhost port number 27017. Optionally, specify a port number to connect a MongoDB instance running on a port other than 27017. --port <port> Species the port number, if the MongoDB instance is not running on the standard port. (i.e. 27017) You may also specify a port number using the mongofiles --host (page 1076) command. --ipv6 Enables IPv6 support that allows mongofiles (page 1075) to connect to the MongoDB instance using an IPv6 network. All MongoDB programs and processes, including mongofiles (page 1075), disable IPv6 support by default. --ssl New in version 2.4: MongoDB added support for SSL connections to mongod (page 1021) instances in mongoles. Note: SSL support in mongoles is not compiled into the default distribution of MongoDB. See Connect to MongoDB with SSL (page 75) for more information on SSL and MongoDB. Additionally, mongoles does not support connections to mongod (page 1021) instances that require client certicate validation. Allows mongofiles (page 1075) to connect to mongod (page 1021) instance over an SSL connection. --username <username>, -u <username> Species a username to authenticate to the MongoDB instance, if your database requires authentication. Use in conjunction with the mongofiles --password (page 1076) option to supply a password. --password <password>, -p <password> Species a password to authenticate to the MongoDB instance. Use in conjunction with the mongofiles --username (page 1076) option to supply a username. If you specify a --username (page 1076) without the --password (page 1076) option, mongofiles (page 1075) will prompt for a password interactively. --authenticationDatabase <dbname> New in version 2.4. Species the database that holds the users (e.g --username) credentials. By default, mongofiles (page 1075) assumes that the database specied to the --db (page 1077) argument holds the users credentials, unless you specify --authenticationDatabase (page 1076). See userSource (page 161), system.users Privilege Documents (page 159) and User Privilege Roles in MongoDB (page 155) for more information about delegated authentication in MongoDB.
1076
--authenticationMechanism <name> New in version 2.4. Species the authentication mechanism. By default, the authentication mechanism is MONGODB-CR, which is the MongoDB challenge/response authentication mechanism. In MongoDB Enterprise, mongofiles (page 1075) also includes support for GSSAPI to handle Kerberos authentication. See Deploy MongoDB with Kerberos Authentication (page 149) for more information about Kerberos authentication. --dbpath <path> Species the directory of the MongoDB data les. If used, the --dbpath (page 1077) option enables mongofiles (page 1075) to attach directly to local data les interact with the GridFS data without the mongod (page 1021). To run with --dbpath (page 1077), mongofiles (page 1075) needs to lock access to the data directory: as a result, no mongod (page 1021) can access the same path while the process runs. --directoryperdb Use the --directoryperdb (page 1077) in conjunction with the corresponding option to mongod (page 1021), which allows mongofiles (page 1075) when running with the --dbpath (page 1077) option and MongoDB uses an on-disk format where every database has a distinct directory. This option is only relevant when specifying the --dbpath (page 1077) option. --journal Allows mongofiles (page 1075) operations to use the durability journal when running with --dbpath (page 1077) to ensure that the database maintains a recoverable state. This forces mongofiles (page 1075) to record all data on disk regularly. --db <db>, -d <db> Use the --db (page 1077) option to specify the MongoDB database that stores or will store the GridFS les. --collection <collection>, -c <collection> This option has no use in this context and a future release may remove it. See SERVER-4931 for more information. --local <filename>, -l <filename> Species the local lesystem name of a le for get and put operations. In the mongoles put and mongoles get commands the required <filename> modier refers to the name the object will have in GridFS. mongofiles (page 1075) assumes that this reects the les name on the local le system. This setting overrides this default. --type <MIME>, t <MIME> Provides the ability to specify a MIME type to describe the le inserted into GridFS storage. mongofiles (page 1075) omits this option in the default operation. Use only with mongoles put operations. --replace, -r Alters the behavior of mongoles put to replace existing GridFS objects with the specied local le, rather than adding an additional object with the same name. In the default operation, les will not be overwritten by a mongoles put option.
Use
To return a list of all les in a GridFS collection in the records database, use the following invocation at the system shell:
1077
mongofiles -d records list
This mongofiles (page 1075) instance will connect to the mongod (page 1021) instance running on the 27017 localhost interface to specify the same operation on a different port or hostname, and issue a command that resembles one of the following:
mongofiles --port 37017 -d records list mongofiles --hostname db1.example.net -d records list mongofiles --hostname db1.example.net --port 37017 -d records list
Modify any of the following commands as needed if youre connecting the mongod (page 1021) instances on different ports or hosts. To upload a le named 32-corinth.lp to the GridFS collection in the records database, you can use the following command:
mongofiles -d records put 32-corinth.lp
To delete the 32-corinth.lp le from this GridFS collection in the records database, you can use the following command:
mongofiles -d records delete 32-corinth.lp
To search for les in the GridFS collection in the records database that have the string corinth in their names, you can use following command:
mongofiles -d records search corinth
To list all les in the GridFS collection in the records database that begin with the string 32, you can use the following command:
mongofiles -d records list 32
To fetch the le from the GridFS collection in the records database named 32-corinth.lp, you can use the following command:
mongofiles -d records get 32-corinth.lp
64.1.7 Run Time Conguration

Conguration File Options
Synopsis
Administrators and users can control mongod (page 1021) or mongos (page 1032) instances at runtime either directly from mongods command line arguments (page 1021) or using a conguration le. While both methods are functionally equivalent and all settings are similar, the conguration le method is preferable. If you installed from a package and have started MongoDB using your systems control script, youre already using a conguration le. To start mongod (page 1021) or mongos (page 1032) using a cong le, use one of the following forms:
mongod mongod mongos mongos --config /etc/mongodb.conf -f /etc/mongodb.conf --config /srv/mongodb/mongos.conf -f /srv/mongodb/mongos.conf
1078
Declare all settings in this le using the following form:

<setting> = <value>
New in version 2.0: Before version 2.0, Boolean (i.e. true|false) or ag parameters, register as true, if they appear in the conguration le, regardless of their value.
Settings
verbose Default: false Increases the amount of internal reporting returned on standard output or in the log le generated by logpath (page 1080). To enable verbose (page 1079) or to enable increased verbosity with vvvv (page 1079), set these options as in the following example:
verbose = true vvvv = true
MongoDB has the following levels of verbosity: v Default: false Alternate form of verbose (page 1079). vv Default: false Additional increase in verbosity of output and logging. vvv Default: false Additional increase in verbosity of output and logging. vvvv Default: false Additional increase in verbosity of output and logging. vvvvv Default: false Additional increase in verbosity of output and logging. port Default: 27017 Species a TCP port for the mongod (page 1021) or mongos (page 1032) instance to listen for client connections. UNIX-like systems require root access for ports with numbers lower than 1024. bind_ip Default: All interfaces. Set this option to congure the mongod (page 1021) or mongos (page 1032) process to bind to and listen for connections from applications on this address. You may attach mongod (page 1021) or mongos (page 1032) instances to any interface; however, if you attach the process to a publicly accessible interface, implement proper authentication or rewall restrictions to protect the integrity of your database. You may concatenate a list of comma separated values to bind mongod (page 1021) to multiple IP addresses.
1079
maxConns Default: depends on system (i.e. ulimit and le descriptor) limits. Unless set, MongoDB will not limit its own connections. Species a value to set the maximum number of simultaneous connections that mongod (page 1021) or mongos (page 1032) will accept. This setting has no effect if it is higher than your operating systems congured maximum connection tracking threshold. This is particularly useful for mongos (page 1032) if you have a client that creates a number of collections but allows them to timeout rather than close the collections. When you set maxConns (page 1079), ensure the value is slightly higher than the size of the connection pool or the total number of connections to prevent erroneous connection spikes from propagating to the members of a shard cluster. Note: You cannot set maxConns (page 1079) to a value higher than 20000. objcheck Default: true Changed in version 2.4: The default setting for objcheck (page 1080) became true in 2.4. In earlier versions objcheck (page 1080) was false by default. Forces the mongod (page 1021) to validate all requests from clients upon receipt to ensure that clients never insert invalid documents into the database. For objects with a high degree of sub-document nesting, objcheck (page 1080) can have a small impact on performance. You can set noobjcheck (page 1080) to disable object checking at run-time. noobjcheck New in version 2.4. Default: false Disables the default object validation that MongoDB performs on all incoming BSON documents. logpath Default: None. (i.e. http://docs.mongodb.org/manual/dev/stdout) Specify the path to a le name for the log le that will hold all diagnostic logging information. Unless specied, mongod (page 1021) will output all log information to the standard output. logappend (page 1080) is true, the logle will be overwritten when the process restarts. Unless
Note: Currently, MongoDB will overwrite the contents of the log le if the logappend (page 1080) is not used. This behavior may change in the future depending on the outcome of SERVER-4499. logappend Default: false Set to true to add new entries to the end of the logle rather than overwriting the content of the log when the process restarts. If this setting is not specied, then MongoDB will overwrite the existing logle upon start up. Note: The behavior of the logging system may change in the near future in response to the SERVER-4499 case. syslog New in version 2.1.0. Sends all logging output to the hosts syslog system rather than to standard output or a log le as with logpath (page 1080).
1080
Warning: You cannot use syslog (page 1080) with logpath (page 1080). pidfilepath Default: None. Specify a le location to hold the PID or process ID of the mongod (page 1021) process. Useful for tracking the mongod (page 1021) process in combination with the fork (page 1081) setting. Without a specied pidfilepath (page 1081), mongos (page 1032) creates no PID le. Without this option, mongod (page 1021) creates no PID le. keyFile Default: None. Specify the path to a key le to store authentication information. This option is only useful for the connection between replica set members. See also: Replica Set Security (page 393) and Replica Set Administration (page 419). nounixsocket Default: false Set to true to disable listening on the UNIX socket. mongod (page 1021) and mongos (page 1032) always listen on the UNIX socket, unless nounixsocket (page 1081) is set, bind_ip (page 1079) is not set, or bind_ip (page 1079) species 127.0.0.1. unixSocketPrefix Default: http://docs.mongodb.org/manual/tmp Species a path for the UNIX socket. Unless this option has a value mongod (page 1021) creates a socket with http://docs.mongodb.org/manual/tmp as a prex. MongoDB will always create and listen on a UNIX socket, unless nounixsocket (page 1081) is set, bind_ip (page 1079) is not set, or bind_ip (page 1079) species 127.0.0.1. fork Default: false Set to true to enable a daemon mode for mongod (page 1021) that runs the process in the background. auth Default: false Set to true to enable database authentication for users connecting from remote hosts. Congure users via the mongo shell (page 1036). If no users exist, the localhost interface will continue to have access to the database until the you create the rst user. cpu Default: false Set to true to force mongod (page 1021) to report every four seconds CPU utilization and the amount of time that the processor waits for I/O operations to complete (i.e. I/O wait.) MongoDB writes this data to standard output, or the logle if using the logpath (page 1080) option. dbpath Default: http://docs.mongodb.org/manual/data/db/
Set this value to designate a directory for the mongod (page 1021) instance to store its data. Typical locations include: http://docs.mongodb.org/manual/srv/mongodb, http://docs.mongodb.org/manual/var/lib/mongodb or http://docs.mongodb.org/manual/opt/mong
1081
Unless specied, mongod (page 1021) will look for data les in the default http://docs.mongodb.org/manual/data/db directory. (Windows systems use the \data\db directory.) If you installed using a package management system. Check the http://docs.mongodb.org/manual/etc/mongodb.conf le provided by your packages to see the conguration of the dbpath (page 1081). diaglog Default: 0 Creates a very verbose, diagnostic log for troubleshooting and recording various errors. MongoDB writes these log les in the dbpath (page 1081) directory in a series of les that begin with the string diaglog with the time logging was initiated appended as a hex string. The value of this setting congures the level of verbosity. Possible values, and their impact are as follows. Value 0 1 2 3 7 Setting off. No logging. Log write operations. Log read operations. Log both read and write operations. Log write and some read operations.
You can use the mongosniff (page 1071) tool to replay this output for investigation. Given a typical diaglog le, located at http://docs.mongodb.org/manual/data/db/diaglog.4f76a58c, you might use a command in the following form to read these les:
mongosniff --source DIAGLOG /data/db/diaglog.4f76a58c
diaglog (page 1082) is for internal use and not intended for most users. Warning: Setting the diagnostic level to 0 will cause mongod (page 1021) to stop writing data to the diagnostic log le. However, the mongod (page 1021) instance will continue to keep the le open, even if it is no longer writing data to the le. If you want to rename, move, or delete the diagnostic log you must cleanly shut down the mongod (page 1021) instance before doing so. directoryperdb Default: false Set to true to modify the storage pattern of the data directory to store each databases les in a distinct folder. This option will create directories within the dbpath (page 1081) named for each directory. Use this option in conjunction with your le system and device conguration so that MongoDB will store data on a number of distinct disk devices to increase write throughput or disk capacity.
1082
Warning: If you have an existing mongod (page 1021) instance and dbpath (page 1081), and you want to enable directoryperdb (page 1082), you must migrate your existing databases to directories before setting directoryperdb (page 1082) to access those databases. Example Given a dbpath (page 1081) directory with the following items:
journal mongod.lock local.0 local.1 local.ns test.0 test.1 test.ns
To enable directoryperdb (page 1082) you would need to modify the dbpath (page 1081) to resemble the following:
journal mongod.lock local/local.0 local/local.1 local/local.ns test/test.0 test/test.1 test/test.ns
journal Default: (on 64-bit systems) true Default: (on 32-bit systems) false Set to true to enable operation journaling to ensure write durability and data consistency. Set to false to prevent the overhead of journaling in situations where durability is not required. To reduce the impact of the journaling on disk usage, you can leave journal (page 1083) enabled, and set smallfiles (page 1085) to true to reduce the size of the data and journal les. Note: You must use nojournal (page 1084) to disable journaling on 64-bit systems. journalCommitInterval Default: 100 or 30 Set this value to specify the maximum amount of time for mongod (page 1021) to allow between journal operations. Lower values increase the durability of the journal, at the possible expense of disk performance. The default journal commit interval is 100 milliseconds if a single block device (e.g. physical volume, RAID device, or LVM volume) contains both the journal and the data les. If different block devices provide the journal and data les the default journal commit interval is 30 milliseconds. This option accepts values between 2 and 300 milliseconds. To force mongod (page 1021) to commit to the journal more frequently, you can specify j:true. When a write operation with j:true is pending, mongod (page 1021) will reduce journalCommitInterval (page 1083) to a third of the set value.
1083
ipv6 Default: false Set to true to IPv6 support to allow clients to connect to mongod (page 1021) using IPv6 networks. mongod (page 1021) disables IPv6 support by default in mongod (page 1021) and all utilities. jsonp Default: false Set to true to permit JSONP access via an HTTP interface. Consider the security implications of allowing this activity before setting this option. noauth Default: true Disable authentication. Currently the default. Exists for future compatibility and clarity. For consistency use the auth (page 1081) option. nohttpinterface Default: false Set to true to disable the HTTP interface. This command will override the rest (page 1085) and disable the HTTP interface if you specify both. Changed in version 2.1.2: The nohttpinterface (page 1084) option is not available for mongos (page 1032) instances before 2.1.2 nojournal Default: (on 64-bit systems) false Default: (on 32-bit systems) true Set nojournal = true to disable durability journaling. By default, mongod (page 1021) enables journaling in 64-bit versions after v2.0. Note: You must use journal (page 1083) to enable journaling on 32-bit systems. noprealloc Default: false Set noprealloc = true to disable the preallocation of data les. This will shorten the start up time in some cases, but can cause signicant performance penalties during normal operations. noscripting Default: false Set noscripting = true to disable the scripting engine. notablescan Default: false Set notablescan = true to forbid operations that require a table scan. nssize Default: 16 Specify this value in megabytes. The maximum size is 2047 megabytes. Use this setting to control the default size for all newly created namespace les (i.e .ns). This option has no impact on the size of existing namespace les. See Limits on namespaces (page 1105).
1084
profile Default: 0 Modify this value to changes the level of database proling, which inserts information about operation performance into output of mongod (page 1021) or the log le if specied by logpath (page 1080). The following levels are available: Level 0 1 2 Setting Off. No proling. On. Only includes slow operations. On. Includes all operations.
By default, mongod (page 1021) disables proling. Database proling can impact database performance because the proler must record and process all database operations. Enable this option only after careful consideration. quota Default: false Set to true to enable a maximum limit for the number data les each database can have. The default quota is 8 data les, when quota is true. Adjust the quota size with the with the quotaFiles (page 1085) setting. quotaFiles Default: 8 Modify limit on the number of data les per database. This option requires the quota (page 1085) setting. rest Default: false Set to true to enable a simple REST interface. repair Default: false Set to true to run a repair routine on all databases following start up. In general you should set this option on the command line and not in the conguration le (page 35) or in a control script. Use the mongod --repair (page 1026) option to access this functionality. Note: Because mongod (page 1021) rewrites all of the database les during the repair routine, if you do not run repair (page 1085) under the same user account as mongod (page 1021) usually runs, you will need to run chown on your database les to correct the permissions before starting mongod (page 1021) again. repairpath Default: A _tmp directory in the dbpath (page 1081). Specify the path to the directory containing MongoDB data les, to use in conjunction with the repair (page 1085) setting or mongod --repair (page 1026) operation. Defaults to a _tmp directory within the dbpath (page 1081). slowms Default: 100 Specify values in milliseconds. Sets the threshold for mongod (page 1021) to consider a query slow for the database proler. The database logs all slow queries to the log, even when the proler is not turned on. When the database proler is on, mongod (page 1021) the proler writes to the system.profile collection. See also: profile (page 1084) 64.1. MongoDB Package Components 1085
smallfiles Default: false Set to true to modify MongoDB to use a smaller default data le size. Specically, smallfiles (page 1085) reduces the initial size for data les and limits them to 512 megabytes. The smallfiles (page 1085) setting also reduces the size of each journal les from 1 gigabyte to 128 megabytes. Use the smallfiles (page 1085) setting if you have a large number of databases that each hold a small quantity of data. The smallfiles (page 1085) setting can lead mongod (page 1021) to create many les, which may affect performance for larger databases. syncdelay Default: 60 mongod (page 1021) writes data very quickly to the journal, and lazily to the data les. syncdelay (page 1086) controls how much time can pass before MongoDB ushes data to the database les via an fsync operation. The default setting is 60 seconds. In almost every situation you should not set this value and use the default setting. The serverStatus (page 889) command reports the background ush threads status via the backgroundFlushing (page 896) eld. syncdelay (page 1086) has no effect on the journal (page 1083) les or journaling (page 71). Warning: If you set syncdelay (page 1086) to 0, MongoDB will not sync the memory mapped les to disk. Do not set this value on production systems. sysinfo Default: false When set to true, mongod (page 1021) returns diagnostic system information regarding the page size, the number of physical pages, and the number of available physical pages to standard output. More typically, run this operation by way of the mongod --sysinfo (page 1028) command. When running with the sysinfo (page 1086), only mongod (page 1021) only outputs the page information and no database process will start. upgrade Default: false When set to true this option upgrades the on-disk data format of the les specied by the dbpath (page 1081) to the latest version, if needed. This option only affects the operation of mongod (page 1021) if the data les are in an old format. When specied for a mongos (page 1032) instance, this option updates the meta data format used by the cong database. Note: In most cases you should not set this value, so you can exercise the most control over your upgrade process. See the MongoDB release notes (on the download page) for more information about the upgrade process. traceExceptions Default: false For internal diagnostic use only. quiet Default: false
1086
Runs the mongod (page 1021) or mongos (page 1032) instance in a quiet mode that attempts to limit the amount of output. This option suppresses: output from database commands, including drop (page 864), dropIndexes (page 864), diagLogging (page 879), validate (page 907), and clean (page 856). replication activity. connection accepted events. connection closed events. Note: For production systems this option is not recommended as it may make tracking problems during particular connections much more difcult. setParameter New in version 2.4. Species an option to congure on startup. Specify multiple options with multiple setParameter (page 1087) options. See mongod Parameters (page 1092) for full documentation of these parameters. The setParameter (page 869) database command provides access to many of these parameters. Declare all setParamater settings in this le using the following form:
setParameter = <parameter>=<value>
For mongod (page 1021) the following options are available using setParameter (page 1087): enableLocalhostAuthBypass (page 1093) enableTestCommands (page 1093) journalCommitInterval (page 1093) logLevel (page 1094) logUserIds (page 1094) notablescan (page 1094) quiet (page 1095) replApplyBatchSize (page 1094) replIndexPrefetch (page 1094) supportCompatibilityFormPrivilegeDocuments (page 1094) syncdelay (page 1094) textSearchEnabled (page 1095) traceExceptions (page 1095) For mongos (page 1032) the following options are available using setParameter (page 1087): enableLocalhostAuthBypass (page 1093) enableTestCommands (page 1093) logLevel (page 1094) logUserIds (page 1094) notablescan (page 1094) quiet (page 1095)
1087
supportCompatibilityFormPrivilegeDocuments (page 1094) syncdelay (page 1094) textSearchEnabled (page 1095) Replication Options replSet Default: <none> Form: <setname> Use this setting to congure replication with replica sets. Specify a replica set name as an argument to this set. All hosts must have the same set name. See also: Replication (page 385), Replica Set Administration (page 419), and Replica Set Conguration (page 463) oplogSize Species a maximum size in megabytes for the replication operation log (e.g. oplog.) mongod (page 1021) creates an oplog based on the maximum amount of space available. For 64-bit systems, the oplog is typically 5% of available disk space. Once the mongod (page 1021) has created the oplog for the rst time, changing oplogSize (page 1088) will not affect the size of the oplog. fastsync Default: false In the context of replica set replication, set this option to true if you have seeded this member with a snapshot of the dbpath of another member of the set. Otherwise the mongod (page 1021) will attempt to perform an initial sync, as though the member were a new member. Warning: If the data is not perfectly synchronized and mongod (page 1021) starts with fastsync (page 1088), then the secondary or slave will be permanently out of sync with the primary, which may cause signicant consistency problems. replIndexPrefetch New in version 2.2. Default: all Values: all, none, and _id_only You can only use replIndexPrefetch (page 1088) in conjunction with replSet (page 1088). By default secondary members of a replica set will load all indexes related to an operation into memory before applying operations from the oplog. You can modify this behavior so that the secondaries will only load the _id index. Specify _id_only or none to prevent the mongod (page 1021) from loading any index into memory. Master/Slave Replication master Default: false Set to true to congure the current instance to act as master instance in a replication conguration. slave Default: false Set to true to congure the current instance to act as slave instance in a replication conguration.
1088
source Default: <> Form: <host><:port> Used with the slave (page 1088) setting to specify the master instance from which this slave instance will replicate only Default: <> Used with the slave (page 1088) option, only (page 1089) species only a single database to replicate. slaveDelay Default: 0 Used with the slave (page 1088) setting, slaveDelay (page 1089) congures a delay in seconds, for this slave to wait to apply operations from the master instance. autoresync Default: false Used with the slave (page 1088) setting, set autoresync (page 1089) to true to force the slave to automatically resync if it is more than 10 seconds behind the master. This setting may be problematic if the oplogSize (page 1088) of the oplog is too small. If the oplog is not large enough to store the difference in changes between the masters current state and the state of the slave, this instance will forcibly resync itself unnecessarily. When you set the autoresync (page 1089) option to false, the slave will not attempt an automatic resync more than once in a ten minute period. Sharded Cluster Options configsvr Default: false Set this value to true to congure this mongod (page 1021) instance to operate as the cong database of a shard cluster. When running with this option, clients will not be able to write data to any database other than config and admin. The default port for a mongod (page 1021) with this option is 27019 and the default dbpath (page 1081) directory is http://docs.mongodb.org/manual/data/configdb, unless specied. Changed in version 2.2: configsvr (page 1089) also sets smallfiles (page 1085). Changed in version 2.4: configsvr (page 1089) creates a local oplog. Do not use configsvr (page 1089) with replSet (page 1088) or shardsvr (page 1089). Cong servers cannot be a shard server or part of a replica set. default port for mongod (page 1021) with this option is 27019 and mongod (page 1021) writes all data les to the http://docs.mongodb.org/manual/configdb sub-directory of the dbpath (page 1081) directory. shardsvr Default: false Set this value to true to congure this mongod (page 1021) instance as a shard in a partitioned cluster. The default port for these instances is 27018. The only affect of shardsvr (page 1089) is to change the port number. configdb Default: None. Format: <config1>,<config2><:port>,<config3>
1089
Set this option to specify a conguration database (i.e. cong database) for the sharded cluster. You must specify either 1 conguration server or 3 conguration servers, in a comma separated list. This setting only affects mongos (page 1032) processes. Note: mongos (page 1032) instances read from the rst cong server in the list provided. All mongos (page 1032) instances must specify the hosts to the configdb (page 1089) setting in the same order. If your conguration databases reside in more that one data center, order the hosts in the configdb (page 1089) setting so that the cong database that is closest to the majority of your mongos (page 1032) instances is rst servers in the list. Warning: Never remove a cong server from the configdb (page 1089) parameter, even if the cong server or servers are not available, or ofine. test Default: false Only runs unit tests and does not start a mongos (page 1032) instance. This setting only affects mongos (page 1032) processes and is for internal testing use only. chunkSize Default: 64 The value of this option determines the size of each chunk of data distributed around the sharded cluster. The default value is 64 megabytes. Larger chunks may lead to an uneven distribution of data, while smaller chunks may lead to frequent and unnecessary migrations. However, in some circumstances it may be necessary to set a different chunk size. This setting only affects mongos (page 1032) processes. Furthermore, chunkSize (page 1090) only sets the chunk size when initializing the cluster for the rst time. If you modify the run-time option later, the new value will have no effect. See the Modify Chunk Size (page 523) procedure if you need to change the chunk size on an existing sharded cluster. localThreshold New in version 2.2. localThreshold (page 1090) affects the logic that program:mongos uses when selecting replica set members to pass reads operations to from clients. Specify a value to localThreshold (page 1090) in milliseconds. The default value is 15, which corresponds to the default value in all of the client drivers (page 555). This setting only affects mongos (page 1032) processes. When mongos (page 1032) receives a request that permits reads to secondary members, the mongos (page 1032) will: nd the member of the set with the lowest ping time. construct a list of replica set members that is within a ping time of 15 milliseconds of the nearest suitable member of the set. If you specify a value for localThreshold (page 1090), mongos (page 1032) will construct the list of replica members that are within the latency allowed by this value. The mongos (page 1032) will select a member to read from at random from this list. The ping time used for a set member compared by the localThreshold (page 1090) setting is a moving average of recent ping times, calculated, at most, every 10 seconds. As a result, some queries may reach members above the threshold until the mongos (page 1032) recalculates the average.
1090
See the Member Selection (page 406) section of the read preference (page 402) documentation for more information. noAutoSplit noAutoSplit (page 1091) is for internal use and is only available on mongos (page 1032) instances. New in version 2.0.7. noAutoSplit (page 1091) prevents mongos (page 1032) from automatically inserting metadata splits in a sharded collection. If set on all mongos (page 1032), this will prevent MongoDB from creating new chunks as the data in a collection grows. Because any mongos (page 1032) in a cluster can create a split, to totally disable splitting in a cluster you must set noAutoSplit (page 1091) on all mongos (page 1032). Warning: With noAutoSplit (page 1091) enabled, the data in your sharded cluster may become imbalanced over time. Enable with caution.
SSL Options sslOnNormalPorts New in version 2.2. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Enables SSL for mongod (page 1021) or mongos (page 1032). With sslOnNormalPorts (page 1091), a mongod (page 1021) or mongos (page 1032) requires SSL encryption for all connections on the default MongoDB port, or the port specied by port (page 1079). By default, sslOnNormalPorts (page 1091) is disabled. sslPEMKeyFile New in version 2.2. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the .pem le that contains both the SSL certicate and key. Specify the le name of the .pem le using relative or absolute paths When using sslOnNormalPorts (page 1091), you must specify sslPEMKeyFile (page 1091). sslPEMKeyPassword New in version 2.2. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the password to de-crypt the certicate-key le (i.e. sslPEMKeyFile (page 1091)). Only use sslPEMKeyPassword (page 1091) if the certicate-key le is encrypted. In all cases, mongod (page 1021) or mongos (page 1032) will redact the password from all logging and reporting output. Changed in version 2.4: sslPEMKeyPassword (page 1091) is only needed when the private key is encrypted. In earlier versions mongod (page 1021) or mongos (page 1032) would require sslPEMKeyPassword (page 1091) whenever using sslOnNormalPorts (page 1091), even when the private key was not encrypted. 64.1. MongoDB Package Components 1091
sslCAFile New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the .pem le that contains the root certicate chain from the Certicate Authority. Specify the le name of the .pem le using relative or absolute paths sslCRLFile New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Species the .pem le that contains the Certicate Revocation List. Specify the le name of the .pem le using relative or absolute paths sslWeakCertificateValidation New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. Disables the requirement for SSL certicate validation, that sslCAFile (page 1091) enables. With sslWeakCertificateValidation (page 1092), mongod (page 1021) or mongos (page 1032) will accept connections if the client does not present a certicate when establishing the connection. If the client presents a certicate and mongod (page 1021) or mongos (page 1032) has sslWeakCertificateValidation (page 1092) enabled, mongod (page 1021) or mongos (page 1032) will validate the certicate using the root certicate chain specied by sslCAFile (page 1091), and reject clients with invalid certicates. Use sslWeakCertificateValidation (page 1092) if you have a mixed deployment that includes clients that do not or cannot present certicates to mongod (page 1021) or mongos (page 1032). sslFIPSMode New in version 2.4. Note: The default distribution of MongoDB does not contain support for SSL. To use SSL you can either compile MongoDB with SSL support or use MongoDB Enterprise. See Connect to MongoDB with SSL (page 75) for more information about SSL and MongoDB. When specied, mongod (page 1021) or mongos (page 1032) will use the FIPS mode of the installed OpenSSL library. Your system must have a FIPS compliant OpenSSL library to use sslFIPSMode (page 1092). mongod Parameters Changed in version 2.4.
1092
Synopsis
MongoDB provides a number of conguration options that are accessible via the --setParameter (page 1027) option to mongod (page 1021). This document documents all of these options. For additional run time conguration options, see Conguration File Options (page 1078) and Manual Page for mongod (page 1021).
Parameters
enableLocalhostAuthBypass New in version 2.4. Specify 0 to disable localhost authentication bypass. Enabled by default. enableLocalhostAuthBypass (page 1093) is not available using setParameter (page 869) database command. Use the setParameter (page 1087) option in the conguration le or the --setParameter (page 1027) option on the command line. enableTestCommands New in version 2.4. enableTestCommands (page 1093) enables a set of internal commands useful for internal testing operations. enableTestCommands (page 1093) is only available when starting mongod (page 1021) and you cannot use setParameter (page 869) to modify this parameter. Consider the following mongod (page 1021) innovation, which sets enableTestCommands (page 1093):
mongod --setParameter enableTestCommands=1
enableTestCommands (page 1093) provides access to the following internal commands: captrunc (page 914) configureFailPoint (page 915) emptycapped (page 915) godinsert (page 916) _hashBSONElement (page 913) journalLatencyTest (page 914) replSetTest (page 916) _skewClockCommand (page 916) sleep (page 916) _testDistLockWithSkew (page 917) _testDistLockWithSyncCluster (page 917) journalCommitInterval Specify an integer between 1 and 500 signifying the number of milliseconds (ms) between journal commits. Consider the following example which sets the journalCommitInterval (page 1093) to 200 ms:
use admin db.runCommand( { setParameter: 1, journalCommitInterval: 200 } )
See also: journalCommitInterval (page 1083). 64.1. MongoDB Package Components 1093
logUserIds New in version 2.4. Specify 1 to enable logging of userids. Disabled by default. logLevel Specify an integer between 0 and 5 signifying the verbosity of the logging, where 5 is the most verbose. Consider the following example which sets the logLevel (page 1094) to 2:
use admin db.runCommand( { setParameter: 1, logLevel: 2 } )
See also: verbose (page 1079). notablescan Specify whether queries must use indexes. If 1, queries that perform a table scan instead of using an index will fail. Consider the following example which sets notablescan (page 1094) to true:
use admin db.runCommand( { setParameter: 1, notablescan: 1 } )
See also: notablescan (page 1084) replIndexPrefetch replApplyBatchSize New in version 2.4. Specify the number of oplog entries to apply as a single batch. replApplyBatchSize (page 1094) must be an integer between 1 and 1024. Only specify this option on secondary members of replica sets. Batch sizes must be 1 for members with slaveDelay (page 1089) congured. saslHostName New in version 2.4. saslHostName (page 1094) overrides MongoDBs default hostname detection for the purpose of conguring SASL and Kerberos authentication. saslHostName (page 1094) does not affect the hostname of the mongod (page 1021) or mongos (page 1032) instance for any purpose beyond the conguration of SASL and Kerberos. You can only set saslHostName (page 1094) during start-up, and cannot change this setting using the setParameter (page 869) database command. Note: saslHostName (page 1094) supports Kerberos authentication and is only included in MongoDB Enterprise. See Deploy MongoDB with Kerberos Authentication (page 149) for more information. supportCompatibilityFormPrivilegeDocuments New in version 2.4. supportCompatibilityFormPrivilegeDocuments (page 1094) is not available using setParameter (page 869) database command. Use the setParameter (page 1087) option in the conguration le or the --setParameter (page 1027) option on the command line.
1094
syncdelay Specify the interval in seconds between fsync operations where mongod (page 1021) ushes its working memory to disk. By default, mongod (page 1021) ushes memory to disk every 60 seconds. Do not change this value unless you see a background ush average greater than 60 seconds. Consider the following example which sets the syncdelay to 60 seconds:
db = db.getSiblingDB("admin") db.runCommand( { setParameter: 1, syncdelay: 60 } )
See also: syncdelay (page 1086). traceExceptions New in version 2.2. Congures mongod (page 1021) log full stack traces on assertions or errors. If 1, mongod (page 1021) will log full stack traces on assertions or errors. Consider the following example which sets the traceExceptions to 1:
use admin db.runCommand( { setParameter: 1, traceExceptions: 1 } )
See also: traceExceptions (page 1086) quiet Sets quiet logging mode. If 1, mongod (page 1021) will go into a quiet logging mode which will not log the following events/activities: connection events; the drop (page 864) command, the dropIndexes (page 864) command, the diagLogging (page 879) command, the validate (page 907) command, and the clean (page 856) command; and replication synchronization activities. Consider the following example which sets the quiet to 1:
db = db.getSiblingDB("admin") db.runCommand( { setParameter: 1, quiet: 1 } )
See also: quiet (page 1086) textSearchEnabled New in version 2.4. Warning: Do not enable or use text search on production systems. Text indexes have signicant storage requirements and performance costs. See text Indexes (page 338) for more information. Enables the text search (page 371) feature. You must enable the feature before creating or accessing a text index.
mongod --setParameter textSearchEnabled=true
If the ag is not enabled, you cannot create new text indexes, and you cannot perform text searches. However, existing text indexes will still be updated. 64.1. MongoDB Package Components 1095
1096
CHAPTER 65
Internal Metadata and Reporting
65.1 The local Database

65.1.1 Overview
Every mongod (page 1021) instance has its own local database, which stores data used in the replication process, and other instance-specic data. The local database is invisible to replication: collections in the local database are not replicated. When running with authentication (i.e. auth (page 1081)), authenticating against the local database is equivalent to authenticating against the admin database. This authentication gives access to all databases. In replication, the local database store stores internal replication data for each member of a replica set. The local stores the following collections:
65.1.2 Collection on all mongod Instances

local.startup_log On startup, each mongod (page 1021) instance inserts a document into startup_log (page 1097) with diagnostic information about the mongod (page 1021) instance itself and host information. startup_log (page 1097) is a capped collection. This information is primarily useful for diagnostic purposes. Example Consider the following prototype of a document from the startup_log (page 1097) collection:
{ "_id" : "<string>", "hostname" : "<string>", "startTime" : ISODate("<date>"), "startTimeLocal" : "<string>", "cmdLine" : { "dbpath" : "<path>", "<option>" : <value> }, "pid" : <number>, "buildinfo" : { "version" : "<string>", "gitVersion" : "<string>", "sysInfo" : "<string>", "loaderFlags" : "<string>",
1097
"compilerFlags" : "<string>", "allocator" : "<string>", "versionArray" : [ <num>, <num>, <...> ], "javascriptEngine" : "<string>", "bits" : <number>, "debug" : <boolean>, "maxBsonObjectSize" : <number> } }
Documents in the startup_log (page 1097) collection contain the following elds: local.startup_log._id Includes the system hostname and a millisecond epoch value. local.startup_log.hostname The systems hostname. local.startup_log.startTime A UTC ISODate value that reects when the server started. local.startup_log.startTimeLocal A string that reports the startTime (page 1098) in the systems local time zone. local.startup_log.cmdLine A sub-document that reports the mongod (page 1021) runtime options and their values. local.startup_log.pid The process identier for this process. local.startup_log.buildinfo A sub-document that reports information about the build environment and settings used to compile this mongod (page 1021). This is the same output as buildInfo (page 871). See buildInfo (page 872).
65.1.3 Collections on Replica Set Members

local.system.replset local.system.replset (page 1098) holds the replica sets conguration object as its single document. To view the objects conguration information, issue rs.conf() (page 988) from the mongo (page 1036) shell. You can also query this collection directly. local.oplog.rs local.oplog.rs (page 1098) is the capped collection that holds the oplog. You set its size at creation using the oplogSize (page 1088) setting. To resize the oplog after replica set initiation, use the Change the Size of the Oplog (page 435) procedure. For additional information, see the Oplog Internals (page 408) section in this document and the Oplog (page 392) section. local.replset.minvalid This contains an object used internally by replica sets to track replication status. local.slaves This contains information about each member of the set and the latest point in time that this member has synced to. If this collection becomes out of date, you can refresh it by dropping the collection and allowing MongoDB to automatically refresh it during normal replication:
db.getSiblingDB("local").slaves.drop()
1098
Chapter 65. Internal Metadata and Reporting
65.1.4 Collections used in Master/Slave Replication

In master/slave replication, the local database contains the following collections: On the master: local.oplog.$main This is the oplog for the master-slave conguration. local.slaves This contains information about each slave. On each slave: local.sources This contains information about the slaves master server.
65.2 System Collections

65.2.1 Synopsis
MongoDB stores system information in collections that use the <database>.system.* namespace, which MongoDB reserves for internal use. Do not create collections that begin with system. MongoDB also stores some additional instance-local metadata in the local database (page 1097), specically for replication purposes.
65.2.2 Collections
System collections include these collections stored directly in the database: <database>.system.namespaces The <database>.system.namespaces (page 1099) collection contains information about all of the databases collections. Additional namespace metadata exists in the database.ns les and is opaque to database users. <database>.system.indexes The <database>.system.indexes (page 1099) collection lists all the indexes in the database. Add and remove data from this collection via the ensureIndex() (page 921) and dropIndex() <database>.system.profile The <database>.system.profile (page 1099) collection stores database proling information. For information on proling, see Database Proling (page 96). <database>.system.users The <database>.system.users (page 160) collection stores credentials for users who have access to the database. For more information on this collection, see Add a User to a Database (page 148) and <database>.system.users (page 160). <database>.system.js The <database>.system.js (page 1099) collection holds special JavaScript code for use in server side JavaScript (page 560). See Store a JavaScript Function on the Server (page 561) for more information.
65.2. System Collections
1099
65.3 Database Proler Output

The database proler captures data information about read and write operations, cursor operations, and database commands. To congure the database prole and set the thresholds for capturing prole data, see the Analyze Performance of Database Operations (page 99) section. The database proler writes data in the system.profile (page 1099) collection, which is a capped collection. To view the prolers output, use normal MongoDB queries on the system.profile (page 1099) collection. Note: Because the database proler writes data to the system.profile (page 1099) collection in a database, the proler will prole some write activity, even for databases that are otherwise read-only.
65.3.1 Example system.profile Document

The documents in the system.profile (page 1099) collection have the following form. This example document reects an update operation:
{ "ts" : ISODate("2012-12-10T19:31:28.977Z"), "op" : "update", "ns" : "social.users", "query" : { "name" : "jane" }, "updateobj" : { "$set" : { "likes" : [ "basketball", "trekking" ] } }, "nscanned" : 8, "moved" : true, "nmoved" : 1, "nupdated" : 1, "keyUpdates" : 0, "numYield" : 0, "lockStats" : { "timeLockedMicros" : { "r" : NumberLong(0), "w" : NumberLong(258) }, "timeAcquiringMicros" : { "r" : NumberLong(0), "w" : NumberLong(7) } }, "millis" : 0, "client" : "127.0.0.1", "user" : "" }
1100
65.3.2 Output Reference

For any single operation, the documents created by the database proler will include a subset of the following elds. The precise selection of elds in these documents depends on the type of operation. system.profile.ts The timestamp of the operation. system.profile.op The type of operation. The possible values are: insert query update remove getmore command system.profile.ns The namespace the operation targets. Namespaces in MongoDB take the form of the database, followed by a dot (.), followed by the name of the collection. system.profile.query The query document used. See Query Specication Documents (page 190) for more information on these documents, and Meta Query Operator Quick Reference (page 801) for more information. system.profile.command The command operation. system.profile.updateobj The update document (page 191) passed in during an update (page 219) operation. system.profile.cursorid The ID of the cursor accessed by a getmore operation. system.profile.ntoreturn Changed in version 2.2: In 2.0, MongoDB includes this eld for query and command operations. In 2.2, this information MongoDB also includes this eld for getmore operations. The number of documents the operation specied to return. For example, the profile (page 888) command would return one document (a results document) so the ntoreturn (page 1101) value would be 1. The limit(5) (page 959) command would return ve documents so the ntoreturn (page 1101) value would be 5. If the ntoreturn (page 1101) value is 0, the command did not specify a number of documents to return, as would be the case with a simple find() (page 924) command with no limit specied. system.profile.ntoskip New in version 2.2. The number of documents the skip() (page 964) method specied to skip. system.profile.nscanned The number of documents that MongoDB scans in the index (page 327) in order to carry out the operation. In general, if nscanned (page 1101) is much higher than nreturned (page 1102), the database is scanning many objects to nd the target objects. Consider creating an index to improve this.
65.3. Database Proler Output
1101
system.profile.moved If moved (page 1101) has a value of true indicates that the update operation moved one or more documents to a new location on disk. These operations take more time than in-place updates, and typically occur when documents grow as a result of document growth. system.profile.nmoved New in version 2.2. The number of documents moved on disk by the operation. system.profile.nupdated New in version 2.2. The number of documents updated by the operation. system.profile.keyUpdates New in version 2.2. The number of index (page 327) keys the update changed in the operation. Changing an index key carries a small performance cost because the database must remove the old key and inserts a new key into the B-tree index. system.profile.numYield New in version 2.2. The number of times the operation yielded to allow other operations to complete. Typically, operations yield when they need access to data that MongoDB has not yet fully read into memory. This allows other operations that have data in memory to complete while MongoDB reads in data for the yielding operation. For more information, see the FAQ on when operations yield (page 728). system.profile.lockStats New in version 2.2. The time in microseconds the operation spent acquiring and holding locks. This eld reports data for the following lock types: R - global read lock W - global write lock r - database-specic read lock w - database-specic write lock system.profile.lockStats.timeLockedMicros The time in microseconds the operation held a specic lock. For operations that require more than one lock, like those that lock the local database to update the oplog, then this value may be longer than the total length of the operation (i.e. millis (page 1102).) system.profile.lockStats.timeAcquiringMicros The time in microseconds the operation spent waiting to acquire a specic lock. system.profile.nreturned The number of documents returned by the operation. system.profile.responseLength The length in bytes of the operations result document. A large responseLength (page 1102) can affect performance. To limit the size of the result document for a query operation, you can use any of the following: Projections (page 171) The limit() method (page 959) The batchSize() method (page 952)
1102
system.profile.millis The time in milliseconds for the server to perform the operation. This time does not include network time nor time to acquire the lock. system.profile.client The IP address or hostname of the client connection where the operation originates. For some operations, such as db.eval() (page 975), the client is 0.0.0.0:0 instead of an actual client. system.profile.user The authenticated user who ran the operation.
65.4 Exit Codes and Statuses

MongoDB will return one of the following codes and statuses when exiting. Use this guide to interpret logs and when troubleshooting issues with mongod (page 1021) and mongos (page 1032) instances. 0 Returned by MongoDB applications upon successful exit. 2 The specied options are in error or are incompatible with other options. 3 Returned by mongod (page 1021) if there is a mismatch between hostnames specied on the command line and in the local.sources (page 1099) collection. mongod (page 1021) may also return this status if oplog collection in the local database is not readable. 4 The version of the database is different from the version supported by the mongod (page 1021) (or mongod.exe (page 1041)) instance. The instance exits cleanly. Restart mongod (page 1021) with the --upgrade (page 1028) option to upgrade the database to the version supported by this mongod (page 1021) instance. 5 Returned by mongod (page 1021) if a moveChunk (page 849) operation fails to conrm a commit. 12 Returned by the mongod.exe (page 1041) process on Windows when it receives a Control-C, Close, Break or Shutdown event. 14 Returned by MongoDB applications which encounter an unrecoverable error, an uncaught exception or uncaught signal. The system exits without performing a clean shut down. 20 Message: ERROR: wsastartup failed <reason> Returned by MongoDB applications on Windows following an error in the WSAStartup function. Message: NT Service Error Returned by MongoDB applications for Windows due to failures installing, starting or removing the NT Service for the application. 45 Returned when a MongoDB application cannot open a le or cannot obtain a lock on a le. 47 MongoDB applications exit cleanly following a large clock skew (32768 milliseconds) event.
65.4. Exit Codes and Statuses
1103
48 mongod (page 1021) exits cleanly if the server socket closes. The server socket is on port 27017 by default, or as specied to the --port (page 1022) run-time option.
49 Returned by mongod.exe (page 1041) or mongos.exe (page 1042) on Windows when either receives a shutdown message from the Windows Service Control Manager. 100 Returned by mongod (page 1021) when the process throws an uncaught exception.
1104
CHAPTER 66
General Reference
66.1 MongoDB Limits and Thresholds

66.1.1 Synopsis
This document provides a collection of hard and soft limitations of the MongoDB system.
66.1.2 Limits
BSON Documents BSON Document Size The maximum BSON document size is 16 megabytes. The maximum document size helps ensure that a single document cannot use excessive amount of RAM or, during transmission, excessive amount of bandwidth. To store documents larger than the maximum size, MongoDB provides the GridFS API. See mongofiles (page 1075) and the documentation for your driver (page 555) for more information about GridFS. Nested Depth for BSON Documents Changed in version 2.2. MongoDB supports no more than 100 levels of nesting for BSON documents. Namespaces Namespace Length Each namespace, including database and collection name, must be shorter than 123 bytes. Number of Namespaces The limitation on the number of namespaces is the size of the namespace le divided by 628. A 16 megabyte namespace le can support approximately 24,000 namespaces. Each index also counts as a namespace. Size of Namespace File Namespace les can be no larger than 2047 megabytes. By default namespace les are 16 megabytes. You can congure the size using the nssize (page 1084) option.
1105
Indexes Index Size Indexed items can be no larger than 1024 bytes. Number of Indexes per Collection A single collection can have no more than 64 indexes. Index Name Length The names of indexes, including their namespace (i.e database and collection name) cannot be longer than 128 characters. The default index name is the concatenation of the eld names and index directions. You can explicitly specify an index name to the ensureIndex() (page 921) helper if the default index name is too long. Unique Indexes in Sharded Collections MongoDB does not support unique indexes across shards, except when the unique index contains the full shard key as a prex of the index. In these situations MongoDB will enforce uniqueness across the full key, not a single eld. See also: Enforce Unique Keys for Sharded Collections (page 533) for an alternate approach. Number of Indexed Fields in a Compound Index There can be no more than 31 elds in a compound index. Capped Collections Maximum Number of Documents in a Capped Collection Changed in version 2.4. If you specify a maximum number of documents for a capped collection using the max parameter to create (page 863), the limit must be less than 232 documents. If you do not specify a maximum number of documents when creating a capped collection, there is no limit on the number of documents. Replica Sets Number of Members of a Replica Set Replica sets can have no more than 12 members. Number of Voting Members of a Replica Set Only 7 members of a replica set can have votes at any given time. See can vote Non-Voting Members (page 389) for more information Sharded Clusters Operations Unavailable in Sharded Environments The group (page 810) does not work with sharding. Use mapReduce (page 814) or aggregate (page 808) instead. db.eval() (page 975) is incompatible with sharded collections. You may use db.eval() (page 975) with un-sharded collections in a shard cluster. $where (page 775) does not permit references to the db object from the $where (page 775) function. This is uncommon in un-sharded collections. The $isolated (page 796) update modier does not work in sharded environments. 1106 Chapter 66. General Reference
$snapshot (page 806) queries do not work in sharded environments. Sharding Existing Collection Data Size MongoDB only allows sharding an existing collection that holds fewer than 256 gigabytes of data. Note: This limitation only applies to sharding collections that have existing data sets, and is not a limit on the size of a sharded collection. See also: Unique Indexes in Sharded Collections (page 1106) Operations Sorted Documents MongoDB will only return sorted results on elds without an index if the sort operation uses less than 32 megabytes of memory. 2d Geospatial queries cannot use the $or operator See also: $or (page 770) and 2d Index Internals (page 369). Spherical Polygons must fit within a hemisphere. Any geometry specied with GeoJSON to $geoIntersects (page 777) or $geoWithin (page 778) queries, must t within a single hemisphere. MongoDB interprets geometries larger than half of the sphere as queries for the smaller of the complementary geometries. Combination Limit with Multiple $in Expressions When using two or more $in (page 765) expressions, the product of the number of distinct elements in the $in (page 765) arrays must be less than 4000000. Otherwise, MongoDB will throw an exception of "combinatorial limit of $in partitioning of result set exceeded". Naming Restrictions Restrictions on Database Names The dot (i.e. .) character is not permissible in database names. Database names are case sensitive even if the underlying le system is case insensitive. Changed in version 2.2: For MongoDB instances running on Windows. In 2.2 the following characters are not permissible in database names:
/\. "*<>:|?
See Restrictions on Database Names for Windows (page 1162) for more information on this change Restriction on Collection Names New in version 2.2. Collection names should begin with an underscore or a letter character, and cannot: contain the $. be an empty string (e.g. ""). contain the null character.
66.1. MongoDB Limits and Thresholds
1107
begin with the system. prex. (Reserved for internal use.) Restrictions on Field Names Field names cannot contain dots (i.e. .), dollar signs (i.e. $), or null characters. See Dollar Sign Operator Escaping (page 718) for an alternate approach.
66.2 Connection String URI Format

This document describes the URI format for dening connections between applications and MongoDB instances in the ofcial MongoDB drivers (page 555).
66.2.1 Standard Connection String Format

This section describes the standard format of the MongoDB connection URI used to connect to a MongoDB database server. The format is the same for all ofcial MongoDB drivers. For a list of drivers and links to driver documentation, see MongoDB Drivers and Client Libraries (page 555). The following is the standard URI connection scheme:
mongodb://[username:password@]host1[:port1][,host2[:port2],...[,hostN[:portN]]][/[database][?options]
The components of this string are: 1. mongodb:// A required prex to identify that this is a string in the standard connection format. 2. username:password@ Optional. If specied, the client will attempt to log in to the specic database using these credentials after connecting to the mongod (page 1021) instance. 3. host1 This the only required part of the URI. It identies a server address to connect to. It identies either a hostname, IP address, or UNIX domain socket. 4. :port1 Optional. The default value is :27017 if not specied. 5. hostX Optional. You can specify as many hosts as necessary. You would specify multiple hosts, for example, for connections to replica sets. 6. :portX Optional. The default value is :27017 if not specied. 7. http://docs.mongodb.org/manual/database Optional. The name of the database to authenticate if the connection string includes authentication credentials in the form of username:password@. If http://docs.mongodb.org/manual/database is not specied and the connection string includes credentials, the driver will authenticate to the admin database. 8. ?options Connection specic options. See Connection String Options (page 1109) for a full description of these options.
1108
Chapter 66. General Reference
If the connection string does not specify a database/ you must specify a slash (i.e. http://docs.mongodb.org/manual/) between the last hostN and the question mark that begins the string of options. Example To describe a connection to a replica set named test, with the following mongod (page 1021) hosts: db1.example.net on port 27017 and db2.example.net on port 2500. You would use a connection string that resembles the following:
mongodb://db1.example.net,db2.example.net:2500/?replicaSet=test
66.2.2 Connection String Options

This section lists all connection options used in the Standard Connection String Format (page 1108).The options are not case-sensitive. Connection options are pairs in the following form: name=value. Separate options with the ampersand (i.e. &) character. In the following example, a connection uses the replicaSet and connectTimeoutMS options:
mongodb://db1.example.net,db2.example.net:2500/?replicaSet=test&connectTimeoutMS=300000
Semi-colon separator for connection string arguments To provide backwards compatibility, drivers currently accept semi-colons (i.e. ;) as option separators.
Replica Set Option uri.replicaSet Species the name of the replica set, if the mongod (page 1021) is a member of a replica set. When connecting to a replica set it is important to give a seed list of at least two mongod (page 1021) instances. If you only provide the connection point of a single mongod (page 1021) instance, and omit the replicaSet (page 1109), the client will create a standalone connection. Connection Options uri.ssl true: Initiate the connection with SSL. false: Initiate the connection without SSL. The default value is false. Note: The ssl (page 1109) option is not supported by all drivers. See your driver (page 555) documentation and the Connect to MongoDB with SSL (page 75) document. uri.connectTimeoutMS The time in milliseconds to attempt a connection before timing out. The default is never to timeout, though different drivers might vary. See the driver (page 555) documentation.
66.2. Connection String URI Format
1109
uri.socketTimeoutMS The time in milliseconds to attempt a send or receive on a socket before the attempt times out. The default is never to timeout, though different drivers might vary. See the driver (page 555) documentation. Connection Pool Options Most drivers implement some kind of connection pooling handle this for you behind the scenes. Some drivers do not support connection pools. See your driver (page 555) documentation for more information on the connection pooling implementation. These options allow applications to congure the connection pool when connecting to the MongoDB deployment. uri.maxPoolSize The maximum number of connections in the connection pool. The default value is 100. uri.minPoolSize The minimum number of connections in the connection pool. The default value is 0. Note: The minPoolSize (page 1110) option is not supported by all drivers. For information on your driver, see the drivers (page 555) documentation. uri.maxIdleTimeMS The maximum number of milliseconds that a connection can remain idle in the pool before being removed and closed. This option is not supported by all drivers. uri.waitQueueMultiple A number that the driver multiples the maxPoolSize (page 1110) value to, to provide the maximum number of threads allowed to wait for a connection to become available from the pool. For default values, see the MongoDB Drivers and Client Libraries (page 555) documentation. uri.waitQueueTimeoutMS The maximum time in milliseconds that a thread can wait for a connection to become available. For default values, see the MongoDB Drivers and Client Libraries (page 555) documentation. Write Concern Options Write concern (page 398) describes the kind of assurances that the program:mongod and the driver provide to the application regarding the success and durability of the write operation. For a full explanation of write concern and write operations in general see the: Write Operations (page 179): uri.w Denes the level and kind of write concern, that the driver uses when calling getLastError (page 831). This option can take either a number or a string as a value. Options -1 The driver will not acknowledge write operations and will suppress all network or socket errors. 0 The driver will not acknowledge write operations, but will pass or handle any network and socket errors that it receives to the client. If you disable write concern but enable the getLastError (page 831) commands journal option, journal overrides this w option. 1 Provides basic acknowledgment of write operations.
1110
By specifying 1, you require that a standalone mongod (page 1021) instance, or the primary for replica sets, acknowledge all write operations. For drivers released after the default write concern change (page 1185), this is the default write concern setting. majority (string) For replica sets, if you specify the special majority value to w (page 1110) option, write operations will only return successfully after a majority of the congured replica set members have acknowledged the write operation. n (number) For replica sets, if you specify a number greater than 1, operations with this write concern will only return after this many members of the set have acknowledged the write. If you set w to a number that is greater than the number of available set members, or members that hold data, MongoDB will wait, potentially indenitely, for these members to become available. tags (string) For replica sets, you can specify a tag set (page 455) to require that all members of the set that have these tags congured return conrmation of the write operation. See Replica Set Tag Set Conguration (page 455) for more information. uri.wtimeoutMS The time in milliseconds to wait for replication to succeed, as specied in the w (page 1110) option, before timing out. uri.journal Controls whether write operations will wait till the mongod (page 1021) acknowledges the write operations and commits the data to the on disk journal. Options true (boolean) Enables journal commit acknowledgment write concern. Equivalent to specifying the getLastError (page 831) command with the j option enabled. false (boolean) Does not require that mongod (page 1021) commit write operations to the journal before acknowledging the write operation. This is the default option for the journal (page 1111) parameter. If you set journal (page 1111) to true, and specify a w (page 1110) value less than 1, journal (page 1111) prevails. If you set journal (page 1111) to true, and the mongod (page 1021) does not have journaling enabled, as with nojournal (page 1084), then getLastError (page 831) will provide basic receipt acknowledgment (i.e. w:1), and will include a jnote eld in its return document. Read Preference Options Read preferences (page 402) describe the behavior of read operations with regards to replica sets. These parameters allow you to specify read preferences on a per-connection basis in the connection string: uri.readPreference Species the replica set read preference for this connection. This setting overrides any slaveOk value. The read preference values are the following: primary (page 403) primaryPreferred (page 403) secondary (page 403) 66.2. Connection String URI Format 1111
secondaryPreferred (page 404) nearest (page 404) For descriptions of each value, see Read Preference Modes (page 402). The default value is primary (page 403), which sends all read operations to the replica sets primary. uri.readPreferenceTags Species a tag set as a comma-separated list of colon-separated key-value pairs. For example:
dc:ny,rack:1
To specify a list of tag sets, use multiple readPreferenceTags. The following species two tag sets and an empty tag set:
readPreferenceTags=dc:ny,rack:1&readPreferenceTags=dc:ny&readPreferenceTags=
Order matters when using multiple readPreferenceTags. Miscellaneous Conguration uri.uuidRepresentation Parameters standard The standard binary representation. csharpLegacy The default representation for the C# driver. javaLegacy The default representation for the Java driver. pythonLegacy The default representation for the Python driver. For the default, see the drivers (page 555) documentation for your driver. Note: Not all drivers support the uuidRepresentation (page 1112) option. For information on your driver, see the drivers (page 555) documentation.
66.2.3 Examples
Consider the following example MongoDB URI strings, that specify common connections: Connect to a database server running locally on the default port:
mongodb://localhost
Connect and log in to the admin database as user sysop with the password moon:
mongodb://sysop:moon@localhost
Connect and log in to the records database as user sysop with the password moon:
mongodb://sysop:moon@localhost/records
Connect to a UNIX domain socket:

mongodb:///tmp/mongodb-27017.sock
1112
Note: Not all drivers support UNIX domain sockets. For information on your driver, see the drivers (page 555) documentation. Connect to a replica set with two members, one on db1.example.net and the other on db2.example.net:
mongodb://db1.example.net,db2.example.com
Connect to a replica set with three members running on localhost, on ports 27017, 27018, and 27019:
mongodb://localhost,localhost:27018,localhost:27019
Connect to a replica set with three members. Send all writes to the primary and distribute reads to the secondaries:
mongodb://example1.com,example2.com,example3.com/?readPreference=secondary
Connect to a replica set with write concern congured to wait for replication to succeed on at least two members, with a two-second timeout.
mongodb://example1.com,example2.com,example3.com/?w=2&wtimeoutMS=2000
66.3 MongoDB Extended JSON

MongoDB import and export utilities (page 105) (i.e. mongoimport (page 1056) and mongoexport (page 1059)) and MongoDB REST Interfaces render an approximation of MongoDB BSON documents in JSON format. The REST interface supports three different modes for document output: Strict mode that produces output that conforms to the JSON RFC specications. JavaScript mode that produces output that most JavaScript interpreters can process (via the --jsonp option) mongo (page 1036) Shell mode produces output that the mongo (page 1036) shell can process. This is extended JavaScript format. MongoDB can process of these representations in REST input. Special representations of BSON data in JSON format make it possible to render information that have no obvious corresponding JSON. In some cases MongoDB supports multiple equivalent representations of the same type information. Consider the following table:
66.3. MongoDB Extended JSON
1113
BSON Data Type data_binary
Strict Mode {
JavaScript (via JSONP)
Mode
mongo Shell Mode
Notes
data_date
data_timestamp
<bindata> is the { BinData ( <t>, <bindata> ) base64 representation "$binary": "<bindata>", "$binary": "<bindata>", of a binary string. "$type": "<t>" "$type": "<t>" <t> is the hexadeci} } mal representation of a single byte that indicates the data type. <date> is the JSON { new Date( <date> n )ew Date ( <date> ) representation of a "$date": <date> 64-bit signed integer } for milliseconds since epoch UTC (unsigned before version 1.9.1). <t> is the JSON rep{ { Timestamp( <t>, <i> ) resentation of a 32-bit "$timestamp": "$timestamp": unsigned integer for { { seconds since epoch. "t": <t>, "t": <t>, <i> is a 32-bit un"i": <i> "i": <i> signed integer for the } } increment. } } {
data_regex
1114 data_oid
<sRegex> is a string /<jRegex>/<jOptions> /<jRegex>/<jOptions> of valid JSON charac"$regex": "<sRegex>", ters. "$options": "<sOptions>" <jRegex> is a } string that may contain valid JSON characters and unescaped double quote (") characters, but may not contain unescaped forward slash (http://docs.mongodb.o characters. <sOptions> is a string containing the regex options represented by the letters of the alphabet. <jOptions> is a string that may contain only the characters g, i, m and s (added in v1.9). Because the JavaScript and mongo Shell representations support a limited range of options, any nonconforming options will be dropped when converting to this representation. Chapter 66. General Reference <id> is a 24{ { ObjectId( "<id>" ) character hexadecimal "$oid": "<id>" "$oid": "<id>" string. } }
66.4 Database References

MongoDB does not support joins. In MongoDB some data is denormalized, or stored with related data in documents to remove the need for joins. However, in some cases it makes sense to store related information in separate documents, typically in different collections or databases. MongoDB applications use one of two methods for relating documents: 1. Manual references (page 1115) where you save the _id eld of one document in another document as a reference. Then your application can run a second query to return the embedded data. These references are simple and sufcient for most use cases. 2. DBRefs (page 1116) are references from one document to another using the value of the rst documents _id eld collection, and optional database name. To resolve DBRefs, your application must perform additional queries to return the referenced documents. Many drivers (page 555) have helper methods that form the query for the DBRef automatically. The drivers 1 do not automatically resolve DBRefs into documents. Use a DBRef when you need to embed documents from multiple collections in documents from one collection. DBRefs also provide a common format and type to represent these relationships among documents. The DBRef format provides common semantics for representing links between documents if your database must interact with multiple frameworks and tools. Unless you have a compelling reason for using a DBRef, use manual references.
66.4.1 Manual References

Background Manual references refers to the practice of including one documents _id eld in another document. The application can then issue a second query to resolve the referenced elds as needed. Process Consider the following operation to insert two documents, using the _id eld of the rst document as a reference in the second document:
original_id = ObjectId() db.places.insert({ "_id": original_id "name": "Broadway Center" "url": "bc.example.net" }) db.people.insert({ "name": "Erin" "places_id": original_id "url": "bc.example.net/Erin" })
Then, when a query returns the document from the people collection you can, if needed, make a second query for the document referenced by the places_id eld in the places collection.
1
Some community supported drivers may have alternate behavior and may resolve a DBRef into a document automatically.
1115
Use For nearly every case where you want to store a relationship between two documents, use manual references (page 1115). The references are simple to create and your application can resolve references as needed. The only limitation of manual linking is that these references do not convey the database and collection name. If you have documents in a single collection that relate to documents in more than one collection, you may need to consider using DBRefs (page 1116).
66.4.2 DBRefs
Background DBRefs are a convention for representing a document, rather than a specic reference type. They include the name of the collection, and in some cases the database, in addition to the value from the _id eld. Format DBRefs have the following elds: $ref The $ref eld holds the name of the collection where the referenced document resides. $id The $id eld contains the value of the _id eld in the referenced document. $db Optional. Contains the name of the database where the referenced document resides. Only some drivers support $db references. Example DBRef document would resemble the following:
{ "$ref" : <value>, "$id" : <value>, "$db" : <value> }
Consider a document from a collection that stored a DBRef in a creator eld:

{ "_id" : ObjectId("5126bbf64aed4daf9e2ab771"), // .. application fields "creator" : { "$ref" : "creators", "$id" : ObjectId("5126bc054aed4daf9e2ab772"), "$db" : "users" } }
The DBRef in this example, points to a document in the creators collection of the users database that has ObjectId("5126bc054aed4daf9e2ab772") in its _id eld. Note: The order of elds in the DBRef matters, and you must use the above sequence when using a DBRef.
1116
Support C++ The C++ driver contains no support for DBRefs. You can transverse references manually. C# The C# driver provides access to DBRef objects with the MongoDBRef Class and supplies the FetchDBRef Method for accessing these objects. Java The DBRef class provides supports for DBRefs from Java. JavaScript The mongo (page 1036) shells JavaScript (page 917) interface provides a DBRef. Perl The Perl driver contains no support for DBRefs. You can transverse references manually or use the MongoDBx::AutoDeref CPAN module. PHP The PHP driver does support DBRefs, including the optional $db reference, through The MongoDBRef class. Python The Python driver provides the DBRef class, and the dereference method for interacting with DBRefs. Ruby The Ruby Driver supports DBRefs using the DBRef class and the deference method. Use In most cases you should use the manual reference (page 1115) method for connecting two or more related documents. However, if you need to reference documents from multiple collections, consider a DBRef.
66.5 GridFS Reference

GridFS stores les in two collections: chunks stores the binary chunks. For details, see The chunks Collection (page 1117). files stores the les metadata. For details, see The les Collection (page 1118). GridFS places the collections in a common bucket by prexing each with the bucket name. By default, GridFS uses two collections with names prexed by fs bucket: fs.files fs.chunks You can choose a different bucket name than fs, and create multiple buckets in a single database. See also: GridFS (page 196) for more information about GridFS.
66.5.1 The chunks Collection

Each document in the chunks collection represents a distinct chunk of a le as represented in the GridFS store. The following is a prototype document from the chunks collection.:
{ "_id" : <string>, "files_id" : <string>, "n" : <num>, "data" : <binary> }
A document from the chunks collection contains the following elds:
66.5. GridFS Reference
1117
chunks._id The unique ObjectID of the chunk. chunks.files_id The _id of the parent document, as specied in the files collection. chunks.n The sequence number of the chunk. GridFS numbers all chunks, starting with 0. chunks.data The chunks payload as a BSON binary type. The chunks collection uses a compound index on files_id and n, as described in GridFS Index (page 197).
66.5.2 The files Collection

Each document in the files collection represents a le in the GridFS store. Consider the following prototype of a document in the files collection:
{ "_id" : <ObjectID>, "length" : <num>, "chunkSize" : <num> "uploadDate" : <timestamp> "md5" : <hash> "filename" : <string>, "contentType" : <string>, "aliases" : <string array>, "metadata" : <dataObject>, }
Documents in the files collection contain some or all of the following elds. Applications may create additional arbitrary elds: files._id The unique ID for this document. The _id is of the data type you chose for the original document. The default type for MongoDB documents is BSON ObjectID. files.length The size of the document in bytes. files.chunkSize The size of each chunk. GridFS divides the document into chunks of the size specied here. The default size is 256 kilobytes. files.uploadDate The date the document was rst stored by GridFS. This value has the Date type. files.md5 An MD5 hash returned from the lemd5 API. This value has the String type. files.filename Optional. A human-readable name for the document. files.contentType Optional. A valid MIME type for the document. files.aliases Optional. An array of alias strings.
1118
files.metadata Optional. Any additional information you want to store.
66.6 Glossary
$cmd A virtual collection that exposes MongoDBs database commands. _id A eld containing a unique ID, typically a BSON ObjectId. If not specied, this value is automatically assigned upon the creation of a new document. You can think of the _id as the documents primary key. accumulator An expression in the aggregation framework that maintains state between documents in the aggregation pipeline. See: $group (page 278) for a list of accumulator operations. admin database A privileged database named admin. Users must have access to this database to run certain administrative commands. See administrative commands (page 121) for more information and Instance Administration Commands (page 855) for a list of these commands. aggregation Any of a variety of operations that reduce and summarize large sets of data. SQLs GROUP and MongoDBs map-reduce are two examples of aggregation functions. aggregation framework The MongoDB aggregation framework provides a means to calculate aggregate values without having to use map-reduce. See also: Aggregation Framework (page 255). arbiter A member of a replica set that exists solely to vote in elections. Arbiters do not replicate data. See also: Delayed Members (page 388) balancer An internal MongoDB process that runs in the context of a sharded cluster and manages the migration of chunks. Administrators must disable the balancer for all maintenance operations on a sharded cluster. BSON A serialization format used to store documents and make remote procedure calls in MongoDB. BSON is a portmanteau of the words binary and JSON. Think of BSON as a binary representation of JSON (JavaScript Object Notation) documents. For a detailed spec, see bsonspec.org. See also: The Data Type Fidelity (page 105) section. BSON types The set of types supported by the BSON serialization format. The following types are available:
66.6. Glossary
1119
Type Double String Object Array Binary data Object id Boolean Date Null Regular Expression JavaScript Symbol JavaScript (with scope) 32-bit integer Timestamp 64-bit integer Min key Max key
Number 1 2 3 4 5 7 8 9 10 11 13 14 15 16 17 18 255 127
btree A data structure used by most database management systems for to store indexes. MongoDB uses b-trees for its indexes. CAP Theorem Given three properties of computing systems, consistency, availability, and partition tolerance, a distributed computing system can provide any two of these features, but never all three. capped collection A xed-sized collection. Once they reach their xed size, capped collections automatically overwrite their oldest entries. MongoDBs oplog replication mechanism depends on capped collections. Developers may also use capped collections in their applications. See also: The Capped Collections (page 558) page. checksum A calculated value used to ensure data integrity. The md5 algorithm is sometimes used as a checksum. chunk In the context of a sharded cluster, a chunk is a contiguous range of shard key values assigned to a particular shard. Chunk ranges are inclusive of the lower boundary and exclusive of the upper boundary. By default, chunks are 64 megabytes or less. When they grow beyond the congured chunk size, a mongos (page 1032) splits the chunk into two chunks. client The application layer that uses a database for data persistence and storage. Drivers provide the interface level between the application layer and the database server. cluster A set of mongod (page 1021) instances running in conjunction to increase database availability and performance. See sharding and replication for more information on the two different approaches to clustering with MongoDB. collection Collections are groupings of BSON documents. Collections do not enforce a schema, but they are otherwise mostly analogous to RDBMS tables. The documents within a collection may not need the exact same set of elds, but typically all documents in a collection have a similar or related purpose for an application. All collections exist within a single database. The namespace within a database for collections are at. See What is a namespace in MongoDB? (page 713) and BSON Documents (page 187) for more information. compound index An index consisting of two or more keys. See Indexing Overview (page 329) for more information.
1120
cong database One of three mongod (page 1021) instances that store all of the metadata associated with a sharded cluster. control script A simple shell script, typically located in the http://docs.mongodb.org/manual/etc/rc.d or http://docs.mongodb.org/manual/etc/init.d directory and used by the systems initialization process to start, restart and stop a daemon process. control script A script used by a UNIX-like operating system to start, stop, or restart a daemon process. On most systems, you can nd these scripts in the http://docs.mongodb.org/manual/etc/init.d/ or http://docs.mongodb.org/manual/etc/rc.d/ directories. CRUD Create, read, update, and delete. The fundamental operations of any database. CSV A text-based data format consisting of comma-separated values. This format is commonly used to exchange database between relational databases, since the format is well-suited to tabular data. You can import CSV les using mongoimport (page 1056). cursor In MongoDB, a cursor is a pointer to the result set of a query, that clients can iterate through to retrieve results. By default, cursors will timeout after 10 minutes of inactivity. daemon The conventional name for a background, non-interactive process. data-center awareness A property that allows clients to address members in a system to based upon their location. Replica sets implement data-center awareness using tagging. See Data Center Awareness (page 61) for more information. database A physical container for collections. Each database gets its own set of les on the le system. A single MongoDB server typically servers multiple databases. database command Any MongoDB operation other than an insert, update, remove, or query. MongoDB exposes commands as queries against the special $cmd collection. For example, the implementation of count (page 808) for MongoDB is a command. See also: Database Commands (page 807) for a full list of database commands in MongoDB database proler A tool that, when enabled, keeps a record on all long-running operations in a databases system.profile collection. The proler is most often used to diagnose slow queries. See also: Monitoring Database Systems (page 96). datum A datum is a set of values used to dene measurements on the earth. MongoDB uses the WGS84 datum. dbpath Refers to the location of MongoDBs data le storage. The default dbpath (page 1081) is http://docs.mongodb.org/manual/data/db. Other common data paths include http://docs.mongodb.org/manual/srv/mongodb and http://docs.mongodb.org/manual/var/lib/mongodb. See also: dbpath (page 1081) or --dbpath (page 1023). delayed member A member of a replica set that cannot become primary and applies operations at a specied delay. This delay is useful for protecting data from human error (i.e. unintentionally deleted databases) or updates that have unforeseen effects on the production database. See also: Delayed Members (page 388)
66.6. Glossary
1121
diagnostic log mongod (page 1021) can create a verbose log of operations with the mongod --diaglog (page 1023) option or through the diagLogging (page 879) command. The mongod (page 1021) creates this log in the directory specied to mongod --dbpath (page 1023). The name of the is diaglog.<time in hex>, where <time-in-hex> reects the initiation time of logging as a hexadecimal string. Warning: Setting the diagnostic level to 0 will cause mongod (page 1021) to stop writing data to the diagnostic log le. However, the mongod (page 1021) instance will continue to keep the le open, even if it is no longer writing data to the le. If you want to rename, move, or delete the diagnostic log you must cleanly shut down the mongod (page 1021) instance before doing so. See also: mongod --diaglog (page 1023), diaglog (page 1082), and diagLogging (page 879). document A record in a MongoDB collection, and the basic unit of data in MongoDB. Documents are analogous to JSON objects, but exist in the database in a more type-rich format known as BSON . dot notation MongoDB uses the dot notation to access the elements of an array and to access the elds of a subdocument. To access an element of an array by the zero-based index position, you concatenate the array name with the dot (.) and zero-based index position:
<array>.<index>
To access a eld of a subdocument with dot-notation, you concatenate the subdocument name with the dot (.) and the eld name:
<subdocument>.<field>
draining The process of removing or shedding chunks from one shard to another. Administrators must drain shards before removing them from the cluster. See also: removeShard (page 850), sharding. driver A client implementing the communication protocol required for talking to a server. The MongoDB drivers provide language-idiomatic methods for interfacing with MongoDB. See also: MongoDB Drivers and Client Libraries (page 555) election In the context of replica sets, an election is the process by which members of a replica set select primaries on startup and in the event of failures. See also: Replica Set Elections (page 389) and priority. eventual consistency A property of a distributed system allowing changes to the system to propagate gradually. In a database system, this means that readable members are not required to reect the latest writes at all times. In MongoDB, reads to a primary have strict consistency; reads to secondaries have eventual consistency. expression In the context of the aggregation framework, expressions are the stateless transformations that operate on the data that passes through the pipeline. See also: Aggregation Framework (page 255).
1122
failover The process that allows one of the secondary members in a replica set to become primary in the event of a failure. See also: Replica Set Failover (page 389). eld A name-value pair in a document. Documents have zero or more elds. Fields are analogous to columns in relational databases. rewall A system level networking lter that restricts access based on, among other things, IP address. Firewalls form part of effective network security strategy. fsync A system call that ushes all dirty, in-memory pages to disk. MongoDB calls fsync() on its database les at least every 60 seconds. geohash A value is a binary representation of the location on a coordinate grid. GeoJSON GeoJSON is a geospatial data interchange format based on JavaScript Object Notation (JSON) and is used in geospatial queries. For supported GeoJSON objects, see Location Data (page 357). For the GeoJSON format specication, see http://geojson.org/geojson-spec.html geospatial Data that relates to geographical location. In MongoDB, you may index or store geospatial data according to geographical parameters and reference specic coordinates in queries. See Geospatial Indexes and Queries (page 357). GridFS A convention for storing large les in a MongoDB database. All of the ofcial MongoDB drivers support this convention, as does the mongofiles program. See also: mongoles (page 1075) and GridFS (page 196). hashed shard key A hashed shard key (page 335) is a special type of shard key that uses a hash of the value in the shard key eld is uses to distribute documents among members of the sharded cluster. haystack index In the context of geospatial queries, haystack indexes enhance searches by creating bucket of objects grouped by a second criterion. For example, you might want all geospatial searches to rst select along a non-geospatial dimension and then match on location. See Haystack Indexes (page 365) for more information. hidden member A member of a replica set that cannot become primary and is not advertised as part of the set in the database command isMaster (page 838), which prevents it from receiving read-only queries depending on read preference. See also: Hidden Member (page 388), isMaster (page 838), db.isMaster local.system.replset.members[n].hidden (page 464). (page 981), and
idempotent When calling an idempotent operation on a value or state, the operation only affects the value once. Thus, the operation can safely run multiple times without unwanted side effects. In the context of MongoDB, oplog entries must be idempotent to support initial synchronization and recovery from certain failure situations. Thus, MongoDB can safely apply oplog entries more than once without any ill effects. index A data structure that optimizes queries. See Indexing Overview (page 329) for more information. initial sync The replica set operation that replicates data from an existing replica set member to a new or restored replica set member. IPv6 A revision to the IP (Internet Protocol) standard that provides a signicantly larger address space to more effectively support the number of hosts on the contemporary Internet. ISODate The international date format used by mongo (page 1036) to display dates. HH:MM.SS.milis. E.g. YYYY-MM-DD
66.6. Glossary
1123
JavaScript A popular scripting language original designed for web browsers. The MongoDB shell and certain server-side functions use a JavaScript interpreter. journal A sequential, binary transaction used to bring the database into a consistent state in the event of a hard shutdown. MongoDB enables journaling by default for 64-bit builds of MongoDB version 2.0 and newer. Journal les are pre-allocated and will exist as three 1GB le in the data directory. To make journal les smaller, use smallfiles (page 1085). When enabled, MongoDB writes data rst to the journal and then to the core data les. MongoDB commits to the journal within 100ms, which is congurable using the journalCommitInterval (page 1083) runtime option. To force mongod (page 1021) to commit to the journal more frequently, you can specify j:true. When a write operation with j:true is pending, mongod (page 1021) will reduce journalCommitInterval (page 1083) to a third of the set value. See also: The Journaling (page 71) page. JSON JavaScript Object Notation. A human-readable, plain text format for expressing structured data with support in many programming languages. JSON document A JSON document is a collection of elds and values in a structured format. The following is a sample JSON document with two elds:
{ name: "MongoDB", type: "database" }
JSONP JSON with Padding. Refers to a method of injecting JSON into applications. Presents potential security concerns. legacy coordinate pairs The format used for geospatial data prior to MongoDB version 2.4. This format stores geospatial data as points on a planar coordinate system. LVM Logical volume manager. LVM is a program that abstracts disk images from physical devices, and provides a number of raw disk manipulation and snapshot capabilities useful for system management. map-reduce A data and processing and aggregation paradigm consisting of a map phase that selects data, and a reduce phase that transforms the data. In MongoDB, you can run arbitrary aggregations over data using map-reduce. See also: The Map-Reduce (page 311) page for more information regarding MongoDBs map-reduce implementation, and Aggregation Framework (page 255) for another approach to data aggregation in MongoDB. master In conventional master/slave replication, the master database receives all writes. The slave instances replicate from the master instance in real time. md5 md5 is a hashing algorithm used to efciently provide reproducible unique strings to identify and checksum data. MongoDB uses md5 to identify chunks of data for GridFS. MIME Multipurpose Internet Mail Extensions. A standard set of type and encoding denitions used to declare the encoding and type of data in multiple data storage, transmission, and email contexts. mongo The MongoDB Shell. mongo connects to mongod (page 1021) and mongos (page 1032) instances, allowing administration, management, and testing. mongo (page 1036) has a JavaScript interface. See also: mongo (page 1036) and mongo Shell Methods (page 917).
1124
mongod The program implementing the MongoDB database server. This server typically runs as a daemon. See also: mongod (page 1021). MongoDB The document-based database server described in this manual. mongos The routing and load balancing process that acts an interface between an application and a MongoDB sharded cluster. See also: mongos (page 1032). multi-master replication A replication method where multiple database instances can accept write operations to the same data set at any time. Multi-master replication exchanges increased concurrency and availability for a relaxed consistency semantic. MongoDB ensures consistency and, therefore, does not provide multi-master replication. namespace The canonical name for a collection or index in MongoDB. The namespace is a combination of the database name and the name of the collection or index, like so: [database-name].[collection-or-index-name]. All documents belong to a namespace. natural order The order in which a database stores documents on disk. Typically, the order of documents on disks reects insertion order, except when documents move internal because of document growth due to update operations. However, Capped collections guarantee that insertion order and natural order are identical. When you execute find() (page 924) with no parameters, the database returns documents in forward natural order. When you execute find() (page 924) and include sort() (page 965) with a parameter of $natural:-1, the database returns documents in reverse natural order. ObjectId A special 12-byte BSON type that has a high probability an ObjectId represent the time of the ObjectIds creation. MongoDB uses ObjectId values as the default values for _id elds. operator A keyword beginning with a $ used to express a complex query, update, or data transformation. For example, $gt is the query languages greater than operator. See the Query, Update and Projection Operators (page 763) for more information about the available operators. oplog A capped collection that stores an ordered history of logical writes to a MongoDB database. The oplog is the basic mechanism enabling replication in MongoDB. See also: Oplog Sizes (page 392) and Change the Size of the Oplog (page 435). ordered query plan Query plan that returns results in the order consistent with the sort() (page 965) order. See also: Query Optimization (page 174) padding The extra space allocated to document on the disk to prevent moving a document when it grows as the result of update() (page 948) operations. padding factor An automatically-calibrated constant used to determine how much extra space MongoDB should allocate per document container on disk. A padding factor of 1 means that MongoDB will allocate only the amount of space needed for the document. A padding factor of 2 means that MongoDB will allocate twice the amount of space required by the document. page fault The event that occurs when a process requests stored data (i.e. a page) from memory that the operating system has moved to disk. See also:
66.6. Glossary
1125
Storage FAQ: What are page faults? (page 746) partition A distributed system architecture that splits data into ranges. Sharding is a kind of partitioning. passive member A member of a replica set that cannot become primary because its priority (page 390) is 0. pcap A packet capture format used by mongosniff (page 1071) to record packets captured from network interfaces and display them as human-readable MongoDB operations. PID A process identier. On UNIX-like systems, a unique integer PID is assigned to each running process. You can use a PID to inspect a running process and send signals to it. pipe A communication channel in UNIX-like systems allowing independent processes to send and receive data. In the UNIX shell, piped operations allow users to direct the output of one command into the input of another. pipeline The series of operations in the aggregation process. See also: Aggregation Framework (page 255). powerOf2Sizes A per-collection setting that changes and normalizes the way that MongoDB allocates space for each document in an effort to maximize storage reuse reduce fragmentation. This is the default for TTL Collections (page 577). See collMod (page 857) and usePowerOf2Sizes (page 857) for more information. New in version 2.2. pre-splitting An operation, performed before inserting data that divides the range of possible shard key values into chunks to facilitate easy insertion and high write throughput. When deploying a sharded cluster, in some cases pre-splitting will expedite the initial distribution of documents among shards by manually dividing the collection into chunks rather than waiting for the MongoDB balancer to create chunks during the course of normal operation. primary In a replica set, the primary member is the current master instance, which receives all write operations. primary key A records unique, immutable identier. In an RDBMS, the primary key is typically an integer stored in each rows id eld. In MongoDB, the _id eld holds a documents primary key which is usually a BSON ObjectId. primary shard For a database where sharding is enabled, the primary shard holds all un-sharded collections. priority In the context of replica sets, priority is a congurable value that helps determine which members in a replica set are most likely to become primary. See also: Replica Set Member Priority (page 390) projection A document given to a query that species which elds MongoDB will return from the documents in the result set. See Projection Operators (page 796) for a list of special projection operators. query A read request. MongoDB queries use a JSON -like query language that includes a variety of query operators with names that begin with a $ character. In the mongo (page 1036) shell, you can issue queries using the db.collection.find() (page 924) and db.collection.findOne() (page 929) methods. query optimizer For each query, the MongoDB query optimizer generates a query plan that matches the query to the index that produces the fastest results. The optimizer then uses the query plan each time the mongod (page 1021) receives the query. If a collection changes signicantly, the optimizer creates a new query plan. See also: Query Optimization (page 174) RDBMS Relational Database Management System. A database management system based on the relational model, typically using SQL as the query language.
1126
read preference A setting on the MongoDB drivers (page 555) that determines how the clients direct read operations. Read preference affects all replica sets including shards. By default, drivers direct all reads to primaries for strict consistency. However, you may also direct reads to secondaries for eventually consistent reads. See also: Read Preference (page 402) read-lock In the context of a reader-writer lock, a lock that while held allows concurrent readers, but no writers. record size The space allocated for a document including the padding. recovering A replica set member status indicating that a member is not ready to begin normal activities of a secondary or primary. Recovering members are unavailable for reads. replica pairs The precursor to the MongoDB replica sets. Deprecated since version 1.6. replica set A cluster of MongoDB servers that implements master-slave replication and automated failover. MongoDBs recommended replication strategy. See also: Replication (page 385) and Replica Set Fundamental Concepts (page 387). replication A feature allowing multiple database servers to share the same data, thereby ensuring redundancy and facilitating load balancing. MongoDB supports two avors of replication: master-slave replication and replica sets. See also: replica set, sharding, Replication (page 385). and Replica Set Fundamental Concepts (page 387). replication lag The length of time between the last operation in the primarys oplog last operation applied to a particular secondary or slave. In general, you want to keep replication lag as small as possible. See also: Replication Lag (page 444) resident memory The subset of an applications memory currently stored in physical RAM. Resident memory is a subset of virtual memory, which includes memory mapped to physical RAM and to disk. REST An API design pattern centered around the idea of resources and the CRUD operations that apply to them. Typically implemented over HTTP. MongoDB provides a simple HTTP REST interface that allows HTTP clients to run commands against the server. rollback A process that, in certain replica set situations, reverts writes operations to ensure the consistency of all replica set members. secondary In a replica set, the secondary members are the current slave instances that replicate the contents of the master database. Secondary members may handle read requests, but only the primary members can handle write operations. secondary index A database index that improves query performance by minimizing the amount of work that the query engine must perform to fulll a query. set name In the context of a replica set, the set name refers to an arbitrary name given to a replica set when its rst congured. All members of a replica set must have the same name specied with the replSet (page 1088) setting (or --replSet (page 1028) option for mongod (page 1021).) See also: replication, Replication (page 385) and Replica Set Fundamental Concepts (page 387).
66.6. Glossary
1127
shard A single mongod (page 1021) instance or a replica set that stores some portion of a sharded clusters total data set. In production, all shards should be replica sets. See sharding. See also: The documents in the Sharding (page 483) section of manual. shard key In a sharded collection, a shard key is the eld that MongoDB uses to distribute documents among members of the sharded cluster. sharded cluster The set of nodes comprising a sharded MongoDB deployment. A sharded cluster consists of three cong processes, one or more replica sets, and one or more mongos (page 1032) routing processes. See also: The documents in the Sharding (page 483) section of manual. sharding A database architecture that enable horizontal scaling by splitting data into key ranges among two or more replica sets. This architecture is also known as range-based partitioning. See shard. See also: The documents in the Sharding (page 483) section of manual. shell helper A number of database commands (page 807) have helper methods in the mongo shell that provide a more concise syntax and improve the general interactive experience. See also: mongo (page 1036) and mongo Shell Methods (page 917). single-master replication A replication topology where only a single database instance accepts writes. Singlemaster replication ensures consistency and is the replication topology employed by MongoDB. slave In conventional master/slave replication, slaves are read-only instances that replicate operations from the master database. Data read from slave instances may not be completely consistent with the master. Therefore, applications requiring consistent reads must read from the master database instance. split The division between chunks in a sharded cluster. SQL Structured Query Language (SQL) is a common special-purpose programming language used for interaction with a relational database including access control as well as inserting, updating, querying, and deleting data. There are some similar elements in the basic SQL syntax supported by different database vendors, but most implementations have their own dialects, data types, and interpretations of proposed SQL standards. Complex SQL is generally not directly portable between major RDBMS products. SQL is often used as metonym for relational databases. SSD Solid State Disk. A high-performance disk drive that uses solid state electronics for persistence, as opposed to the rotating platters and movable read/write heads used by traditional mechanical hard drives. standalone In MongoDB, a standalone is an instance of mongod (page 1021) that is running as a single server and not as part of a replica set. strict consistency A property of a distributed system requiring that all members always reect the latest changes to the system. In a database system, this means that any system that can provide data must reect the latest writes at all times. In MongoDB, reads to a primary have strict consistency; reads to secondary members have eventual consistency. sync The replica set operation where members replicate data from the primary. Replica sets synchronize data at two different points: Initial sync occurs when MongoDB creates new databases on a new or restored replica set member, populating the member with the replica sets data.
1128
Replication occurs continually after initial sync and keeps the member updated with changes to the replica sets data. syslog On UNIX-like systems, a logging process that provides a uniform standard for servers and processes to submit logging information. tag One or more labels applied to a given replica set member that clients may use to issue data-center aware operations. TSV A text-based data format consisting of tab-separated values. This format is commonly used to exchange database between relational databases, since the format is well-suited to tabular data. You can import TSV les using mongoimport (page 1056). TTL Stands for time to live, and represents an expiration time or period for a given piece of information to remain in a cache or other temporary storage system before the system deletes it or ages it out. unique index An index that enforces uniqueness for a particular eld across a single collection. unordered query plan Query plan that returns results in an order inconsistent with the sort() (page 965) order. See also: Query Optimization (page 174) upsert A kind of update that either updates the rst document matched in the provided query selector or, if no document matches, inserts a new document having the elds implied by the query selector and the update operation. virtual memory An applications working memory, typically residing on both disk an in physical RAM. WGS84 The default datum MongoDB uses to calculate geometry over an Earth-like sphere. MongoDB uses the WGS84 datum for geospatial queries on GeoJSON objects. See http://spatialreference.org/ref/epsg/4326/. working set The collection of data that MongoDB uses regularly. This data is typically (or preferably) held in RAM. write concern Species whether a write operation has succeeded. Write concern allows your application to detect insertion errors or unavailable mongod (page 1021) instances. For replica sets, you can congure write concern to conrm replication to a specied number of members. See also: Write Concern (page 398), Write Operations (page 179), and Write Concern for Replica Sets (page 398). write-lock A lock on the database for a given writer. When a process writes to the database, it takes an exclusive write-lock to prevent other processes from writing or reading. writeBacks The process within the sharding system that ensures that writes issued to a shard that isnt responsible for the relevant chunk, get applied to the proper shard. See also: The genindex may provide useful insight into the reference material in this manual.
66.6. Glossary
1129
1130
Part XV
Release Notes
1131
Always install the latest, stable version of MongoDB. See Version Numbers (page 1187) for more information. See the following release notes for an account of the changes in major versions. Release notes also include instructions for upgrade.
1133
1134
CHAPTER 67
Current Stable Release
(2.4-series)
67.1 Release Notes for MongoDB 2.4

See the full index of this page for a complete list of changes included in 2.4. Platform Support (page 1137) Upgrade Process (page 1137) Changes (page 1144) Major Features (page 1144) Security Improvements (page 1145) Administration Changes (page 1146) Indexing Changes (page 1148) Interface Changes (page 1148) Additional Resources (page 1155) MongoDB 2.4 was released on March 19, 2013.
67.1.1 Whats New in MongoDB 2.4

MongoDB 2.4 represents hundreds of improvements and features driven by user requests. MongoDB 2.4 builds on the momentum of 2.2 by introducing new features that enable greater developer productivity, easier operations, improved performance and enhanced security. MongoDB 2.4 is available for download on MongoDB.org. Developer Productivity Aggregation Framework renements include an overhaul of the underlying engine introduced in MongoDB 2.2 making it easier to leverage real-time, in-place analytics. MongoDB 2.4 includes signicant performance improvements, additional support for binary data, support for $geoWithin (page 778) and $near (page 780) geospatial queries, improved string concatenation with the new $concat (page 272) operator, and improved date calculation semantics. Geospatial enhancements support new use cases with support for polygon intersection queries (with $geoIntersects (page 777)), support for GeoJSON , and an improved spherical model. Learn more about geospatial improvements in 2 (page 1145).
1135
Text Search provides a simplied, integrated approach to incorporating search functionality into apps with support for language specic stemming and stop words in 15 languages and real time indexes. Text search is beta for 2.4 and is not recommended for production use. Learn more about text search (page 1144). New update semantics for arrays with the $push (page 792) update operator. Applications can now use $slice (page 793) to maintain xed size arrays, and use $sort (page 794) to maintain sorted arrays. Learn more about capped arrays (page 1148). New $setOnInsert (page 787) update operator supports specifying elds to add only on insert and upsert operations. Ease of Operations Hashed indexes and shard keys provide simple, even distribution for reads and writes. Learn more about hashed indexes and shard keys (page 1145). New serverStatus (page 889) metrics including a working set analysis tool makes capacity planning easier for operations teams. Learn more about the new serverStatus metrics (page 1147) including the working set analyzer (page 902). More control for operators with the ability to terminate indexing operations with automatic resource cleanup. Improved Performance V8 JavaScript engine offers better performance and concurrency with JavaScript based actions including those using the $where (page 775) query operator as well as mapReduce (page 814) and eval (page 823). Learn more about MongoDB on V8 (page 1149), and JavaScript Changes in MongoDB 2.4 (page 1149). Improvements to count (page 808) provide dramatically faster count operations. Counting is now up to 20 times faster for low cardinality index based counts. Signicant optimizations to $elemMatch (page 783) when using a multi-key index. More Robust Security Role-Based privileges allow organizations to assign more granular security policies for server, database and cluster administration. Learn more about role based access control in MongoDB (page 155). Kerberos authentication mechanism in MongoDB Enterprise. MongoDB Enterprise MongoDB Enterprise is a commercial edition of MongoDB that includes enterprise-grade capabilities, such as advanced security features, management tools, software integrations and certications. Available as part of the MongoDB Enterprise Subscription, this edition includes 10gens most comprehensive SLA and a commercial license. Continue reading for more information on MongoDB Enterprise Learning More These features represent only a small portion of the improvements made in MongoDB 2.4. For more details see the MongoDB 2.4 release notes (page 1135) and Jira for a complete list of all cases closed for MongoDB 2.4 sorted by user votes
1136
Chapter 67. Current Stable Release
67.1.2 Platform Support

For OS X, MongoDB 2.4 only supports OS X versions 10.6 (Snow Leopard) and later. There are no other platform support changes in MongoDB 2.4. See the downloads page for more information on platform support.
67.1.3 Upgrade Process

Upgrade MongoDB to 2.4 In the general case, the upgrade from MongoDB 2.2 to 2.4 is a binary-compatible drop-in upgrade: shut down the mongod (page 1021) instances and replace them with mongod (page 1021) instances running 2.4. However, before you attempt any upgrade please familiarize yourself with the content of this document, particularly the procedure for upgrading sharded clusters (page 1138) and the considerations for reverting to 2.2 after running 2.4 (page 1142). Content Upgrade Recommendations and Checklist (page 1137) Upgrade Standalone mongod Instance to MongoDB 2.4 (page 1137) Upgrade a Replica Set from MongoDB 2.2 to MongoDB 2.4 (page 1138) Upgrade a Sharded Cluster from MongoDB 2.2 to MongoDB 2.4 (page 1138) Rolling Upgrade Limitation for 2.2.0 Deployments Running with auth Enabled (page 1142) Upgrade from 2.3 to 2.4 (page 1142) Downgrade MongoDB from 2.4 to Previous Versions (page 1142)
Upgrade Recommendations and Checklist
When upgrading, consider the following: For all deployments using authentication, upgrade the drivers (i.e. client libraries), before upgrading the mongod (page 1021) instance or instances. To upgrade to 2.4 sharded clusters must upgrade following the meta-data upgrade procedure (page 1138). If youre using 2.2.0 and running with auth (page 1081) enabled, you will need to upgrade rst to 2.2.1 and then upgrade to 2.4. See Rolling Upgrade Limitation for 2.2.0 Deployments Running with auth Enabled (page 1142). If you have system.users (page 160) documents (i.e. for auth (page 1081)) that you created before 2.4 you must ensure that there are no duplicate values for the user eld in the system.users (page 160) collection in any database. If you do have documents with duplicate user elds, you must remove them before upgrading. See Compatibility Change: User Uniqueness Enforced (page 1146) for more information.
Upgrade Standalone mongod Instance to MongoDB 2.4
1. Download binaries of the latest release in the 2.4 series from the MongoDB Download Page. See Install MongoDB (page 3) for more information. 2. Shutdown your mongod (page 1021) instance. Replace the existing binary with the 2.4 mongod (page 1021) binary and restart mongod (page 1021).
67.1. Release Notes for MongoDB 2.4
1137
Upgrade a Replica Set from MongoDB 2.2 to MongoDB 2.4
You can upgrade to 2.4 by performing a rolling upgrade of the set by upgrading the members individually while the other members are available to minimize downtime. Use the following procedure: 1. Upgrade the secondary members of the set one at a time by shutting down the mongod (page 1021) and replacing the 2.2 binary with the 2.4 binary. After upgrading a mongod (page 1021) instance, wait for the member to recover to SECONDARY state before upgrading the next instance. To check the members state, issue rs.status() (page 990) in the mongo (page 1036) shell. 2. Use the mongo (page 1036) shell method rs.stepDown() (page 991) to step down the primary to allow the normal failover (page 389) procedure. rs.stepDown() (page 991) expedites the failover procedure and is preferable to shutting down the primary directly. Once the primary has stepped down and another member has assumed PRIMARY state, as observed in the output of rs.status() (page 990), shut down the previous primary and replace mongod (page 1021) binary with the 2.4 binary and start the new process. Note: Replica set failover is not instant but will render the set unavailable to read or accept writes until the failover process completes. Typically this takes 10 seconds or more. You may wish to plan the upgrade during a predened maintenance window.
Upgrade a Sharded Cluster from MongoDB 2.2 to MongoDB 2.4
Important: Only upgrade sharded clusters to 2.4 if all members of the cluster are currently running instances of 2.2. The only supported upgrade path for sharded clusters running 2.0 is via 2.2. Upgrading a sharded cluster from MongoDB version 2.2 to 2.4 (or 2.3) requires that you run a 2.4 mongos (page 1032) with the --upgrade (page 1034) option, described in this procedure. The upgrade process does not require downtime. The upgrade to MongoDB 2.4 adds epochs to the meta-data for all collections and chunks in the existing cluster. MongoDB 2.2 processes are capable of handling epochs, even though 2.2 did not require them. This procedure applies only to upgrades from version 2.2. Earlier versions of MongoDB do not correctly handle epochs.
1138
Warning: Before you start the upgrade, ensure that the amount of free space on the lesystem for the cong database (page 545) is 4 to 5 times the amount of space currently used by the cong database (page 545) data les. While the upgrade is in progress, you cannot make changes to the collection meta-data. For example, during the upgrade, do not perform: sh.enableSharding() (page 998), sh.shardCollection() (page 1000), sh.addShard() (page 995), db.createCollection() (page 970), db.collection.drop() (page 921), db.dropDatabase() (page 975), any operation that creates a database, or any other operation that modies the cluster meta-data in any way. See Sharding Commands (page 537) for a complete list of sharding commands. Note, however, that not all commands on the Sharding Commands (page 537) page modies the cluster meta-data. Once you upgrade to 2.4 and complete the upgrade procedure do not use 2.0 mongod (page 1021) and mongos (page 1032) processes in your cluster. 2.0 process may re-introduce old meta-data formats into cluster meta-data. Note: The upgraded cong database will require more storage space than before, to make backup and working copies of the config.chunks (page 547) and config.collections (page 548) collections. As always, if storage requirements increase, the mongod (page 1021) might need to pre-allocate additional data les. See What tools can I use to investigate storage use in MongoDB? (page 746) for more information.
Meta-Data Upgrade Procedure Changes to the meta-data format for sharded clusters, stored in the cong database (page 545), require a special meta-data upgrade procedure when moving to 2.4. Do not perform operations that modify meta-data while performing this procedure. See Upgrade a Sharded Cluster from MongoDB 2.2 to MongoDB 2.4 (page 1138) for examples of prohibited operations. 1. Before you start the upgrade, ensure that the amount of free space on the lesystem for the cong database (page 545) is 4 to 5 times the amount of space currently used by the cong database (page 545) data les. 2. Turn off the balancer (page 497) in the sharded cluster, as described in Disable the Balancer (page 529). Optional For additional security during the upgrade, you can make a backup of the cong database using mongodump (page 1044) or other backup tools. 3. Ensure there are no version 2.0 mongod (page 1021) or mongos (page 1032) processes still active in the sharded cluster. The automated upgrade process checks for 2.0 processes, but network availability can prevent a denitive check. Wait 5 minutes after stopping or upgrading version 2.0 mongos (page 1032) processes to conrm that none are still active. 4. Start a single 2.4 mongos (page 1032) process with configdb (page 1089) pointing to the sharded clusters cong servers (page 500) and with the --upgrade (page 1034) option. The upgrade process happens before the process becomes a daemon (i.e. before --fork (page 1034).) You can upgrade an existing mongos (page 1032) instance to 2.4 or you can start a new mongos instance that can reach all cong servers if you need to avoid reconguring a production mongos (page 1032). Start the mongos (page 1032) with a command that resembles the following:
1139
mongos --configdb <config server> --upgrade
Without the --upgrade (page 1034) option 2.4 mongos (page 1032) processes will fail to start until the upgrade process is complete. The upgrade will prevent any chunk moves or splits from occurring during the upgrade process. If there are very many sharded collections or there are stale locks held by other failed processes, acquiring the locks for all collections can take seconds or minutes. See the log for progress updates. 5. When the mongos (page 1032) process starts successfully, the upgrade is complete. If the mongos (page 1032) process fails to start, check the log for more information. If the mongos (page 1032) terminates or loses its connection to the cong servers during the upgrade, you may always safely retry the upgrade. However, if the upgrade failed during the short critical section, the mongos (page 1032) will exit and report that the upgrade will require manual intervention. To continue the upgrade process, you must follow the Resync after an Interruption of the Critical Section (page 1140) procedure. Optional If the mongos (page 1032) logs show the upgrade waiting for the upgrade lock, a previous upgrade process may still be active or may have ended abnormally. After 15 minutes of no remote activity mongos (page 1032) will force the upgrade lock. If you can verify that there are no running upgrade processes, you may connect to a 2.2 mongos (page 1032) process and force the lock manually:
mongo <mongos.example.net> db.getMongo().getCollection("config.locks").findOne({ _id : "configUpgrade" })
If the process specied in the process eld of this document is veriably ofine, run the following operation to force the lock.
db.getMongo().getCollection("config.locks").update({ _id : "configUpgrade" }, { $set : { state :
It is always more safe to wait for the mongos (page 1032) to verify that the lock is inactive, if you have any doubts about the activity of another upgrade operation. In addition to the configUpgrade, the mongos (page 1032) may need to wait for specic collection locks. Do not force the specic collection locks. 6. Upgrade and restart other mongos (page 1032) processes in the sharded cluster, without the --upgrade (page 1034) option. See Complete Sharded Cluster Upgrade (page 1141) for more information. 7. Re-enable the balancer (page 529). You can now perform operations that modify cluster meta-data. Once you have upgraded, do not introduce version 2.0 MongoDB processes into the sharded cluster. This can reintroduce old meta-data formats into the cong servers. The meta-data change made by this upgrade process will help prevent errors caused by cross-version incompatibilities in future versions of MongoDB. Resync after an Interruption of the Critical Section During the short critical section of the upgrade that applies changes to the meta-data, it is unlikely but possible that a network interruption can prevent all three cong servers from verifying or modifying data. If this occurs, the cong servers (page 500) must be re-synced, and there may be problems starting new mongos (page 1032) processes. The sharded cluster will remain accessible, but avoid all cluster meta-data changes until you resync the cong servers. Operations that change meta-data include: adding shards, dropping databases, and dropping collections. Note: Only perform the following procedure if something (e.g. network, power, etc.) interrupts the upgrade process
1140
during the short critical section of the upgrade. Remember, you may always safely attempt the meta data upgrade procedure (page 1139). To resync the cong servers: 1. Turn off the balancer (page 497) in the sharded cluster and stop all meta-data operations. If you are in the middle of an upgrade process (Upgrade a Sharded Cluster from MongoDB 2.2 to MongoDB 2.4 (page 1138)), you have already disabled the balancer. 2. Shut down two of the three cong servers, preferably the last two listed in the configdb (page 1089) string. For example, if your configdb (page 1089) string is configA:27019,configB:27019,configC:27019, shut down configB and configC. Shutting down the last two cong servers ensures that most mongos (page 1032) instances will have uninterrupted access to cluster meta-data. 3. mongodump (page 1044) the data les of the active cong server (configA). 4. Move the data les of the deactivated cong servers (configB and configC) to a backup location. 5. Create new, empty data directories. 6. Restart the disabled cong servers with --dbpath (page 1023) pointing to the now-empty data directory and --port (page 1022) pointing to an alternate port (e.g. 27020). 7. Use mongorestore (page 1048) to repopulate the data les on the disabled documents from the active cong server (configA) to the restarted cong servers on the new port (configB:27020,configC:27020). These cong servers are now re-synced. 8. Restart the restored cong servers on the old port, resetting the port back to the old settings (configB:27019 and configC:27019). 9. In some cases connection pooling may cause spurious failures, as the mongos (page 1032) disables old connections only after attempted use. 2.4 xes this problem, but to avoid this issue in version 2.2, you can restart all mongos (page 1032) instances (one-by-one, to avoid downtime) and use the rs.stepDown() (page 991) method before restarting each of the shard replica set primaries. 10. The sharded cluster is now fully resynced; however before you attempt the upgrade process again, you must manually reset the upgrade state using a version 2.2 mongos (page 1032). Begin by connecting to the 2.2 mongos (page 1032) with the mongo (page 1036) shell:
mongo <mongos.example.net>
Then, use the following operation to reset the upgrade process:

db.getMongo().getCollection("config.version").update({ _id : 1 }, { $unset : { upgradeState : 1
11. Finally retry the upgrade process, as in Upgrade a Sharded Cluster from MongoDB 2.2 to MongoDB 2.4 (page 1138). Complete Sharded Cluster Upgrade After you have successfully completed the meta-data upgrade process described in Meta-Data Upgrade Procedure (page 1139), and the 2.4 mongos (page 1032) instance starts, you can upgrade the other processes in your MongoDB deployment. While the balancer is still disabled, upgrade the components of your sharded cluster in the following order: Upgrade all mongos (page 1032) instances in the cluster, in any order. Upgrade all 3 mongod (page 1021) cong server instances, upgrading the rst system in the mongos --configdb (page 1034) argument last.
1141
Upgrade each shard, one at a time, upgrading the mongod (page 1021) secondaries before running replSetStepDown (page 844) and upgrading the primary of each shard. When this process is complete, you can now re-enable the balancer (page 529).
Rolling Upgrade Limitation for 2.2.0 Deployments Running with auth Enabled
MongoDB cannot support deployments that mix 2.2.0 and 2.4.0, or greater, components. MongoDB version 2.2.1 and later processes can exist in mixed deployments with 2.4-series processes. Therefore you cannot perform a rolling upgrade from MongoDB 2.2.0 to MongoDB 2.4.0. To upgrade a cluster with 2.2.0 components, use one of the following procedures. 1. Perform a rolling upgrade of all 2.2.0 processes to the latest 2.2-series release (e.g. 2.2.3) so that there are no processes in the deployment that predate 2.2.1. When there are no 2.2.0 processes in the deployment, perform a rolling upgrade to 2.4.0. 2. Stop all processes in the cluster. Upgrade all processes to a 2.4-series release of MongoDB, and start all processes at the same time.
Upgrade from 2.3 to 2.4
If you used a mongod (page 1021) from the 2.3 or 2.4-rc (release candidate) series, you can safely transition these databases to 2.4.0 or later; however, if you created 2dsphere or text indexes using a mongod (page 1021) before v2.4-rc2, you will need to rebuild these indexes. For example:
db.records.dropIndex( { loc: "2dsphere" } ) db.records.dropIndex( "records_text" ) db.records.ensureIndex( { loc: "2dsphere" } ) db.records.ensureIndex( { records: "text" } )
Downgrade MongoDB from 2.4 to Previous Versions
For some cases the on-disk format of data les used by 2.4 and 2.2 mongod (page 1021) is compatible, and you can upgrade and downgrade if needed. However, several new features in 2.4 are incompatible with previous versions: 2dsphere indexes are incompatible with 2.2 and earlier mongod (page 1021) instances. text indexes are incompatible with 2.2 and earlier mongod (page 1021) instances. using a hashed index as a shard key are incompatible with 2.2 and earlier mongos (page 1032) instances hashed indexes are incompatible with 2.0 and earlier mongod (page 1021) instances. Note: In sharded clusters, once you have completed the meta-data upgrade procedure (page 1138), you cannot use 2.0 mongod (page 1021) or mongos (page 1032) instances in the same cluster. If you complete the meta-data upgrade, you can have a mixed cluster that has both 2.2 and 2.4 mongod (page 1021) and mongos (page 1032) instances, if needed. However, do not create 2dsphere or text indexes in a cluster that has 2.2 components.
Considerations and Compatibility If you upgrade to MongoDB 2.4, and then need to run MongoDB 2.2 with the same data les, consider the following limitations.
1142
If you use a hashed index as the shard key index, which is only possible under 2.4 you will not be able to query data in this sharded collection. Furthermore, a 2.2 mongos (page 1032) cannot properly route an insert operation for a collections sharded using a hashed index for the shard key index: any data that you insert using a 2.2 mongos (page 1032), will not arrive on the correct shard and will not be reachable by future queries. If you never create an 2dsphere or text index, you can move between a 2.4 and 2.2 mongod (page 1021) for a given data set; however, after you create the rst 2dsphere or text index with a 2.4 mongod (page 1021) you will need to run a 2.2 mongod (page 1021) with the --upgrade (page 1028) option and drop any 2dsphere or text index. Upgrade and Downgrade Procedures Basic Downgrade and Upgrade Except as described below, moving between 2.2 and 2.4 is a drop-in replacement:
stop the existing mongod (page 1021), using the --shutdown (page 1028) option as follows:
mongod --dbpath /var/mongod/data --shutdown
Replace http://docs.mongodb.org/manual/var/mongod/data with your MongoDB dbpath (page 1081). start the new mongod (page 1021) processes with the same dbpath (page 1081) setting, for example:
mongod --dbpath /var/mongod/data
Replace http://docs.mongodb.org/manual/var/mongod/data with your MongoDB dbpath (page 1081). Downgrade to 2.2 After Creating a 2dsphere or text Index If you have created 2dsphere or text indexes while running a 2.4 mongod (page 1021) instance, you can downgrade at any time, by starting the 2.2 mongod (page 1021) with the --upgrade (page 1028) option as follows:
mongod --dbpath /var/mongod/data/ --upgrade
Then, you will need to drop any existing 2dsphere or text indexes using db.collection.dropIndex(), for example:
db.records.dropIndex( { loc: "2dsphere" } ) db.records.dropIndex( "records_text" )
Warning: --upgrade (page 1028) will run repairDatabase (page 868) on any database where you have created a 2dsphere or text index, which will rebuild all indexes.
Troubleshooting Upgrade/Downgrade Operations If you do not use --upgrade (page 1028), when you attempt to start a 2.2 mongod (page 1021) and you have created a 2dsphere or text index, mongod (page 1021) will return the following message:
need to upgrade database index_plugin_upgrade with pdfile version 4.6, new version: 4.5 Not upgradin
While running 2.4, to check the data le version of a MongoDB database, use the following operation in the shell:
db.getSiblingDB(<databasename>).stats().dataFileVersion
The major data le version for both 2.2 and 2.4 is 4, the minor data le version for 2.2 is 5 and the minor data le version for 2.4 is 6 if you have created a 2dsphere or text.
1143
Compatibility and Index Type Changes in MongoDB 2.4 In 2.4 MongoDB includes two new features related to indexes that users upgrading to version 2.4 must consider, particularly with regard to possible downgrade paths. For more information on downgrades, see Downgrade MongoDB from 2.4 to Previous Versions (page 1142).
New Index Types
In 2.4 MongoDB adds two new index types: 2dsphere and text. These index types do not exist in 2.2, and for each database, creating a 2dsphere or text index, will upgrade the data-le version and make that database incompatible with 2.2. If you intend to downgrade, you should always drop all 2dsphere and text indexes before moving to 2.2. You can use the downgrade procedure (page 1142) to downgrade these databases and run 2.2 if needed, however this will run a full database repair (as with repairDatabase (page 868),) for all affected databases.
Index Type Validation
In MongoDB 2.2 and earlier you could specify invalid index types that did not exist. In these situations, MongoDB would create an ascending (e.g. 1) index. Invalid indexes include index types specied by strings that do not refer to an existing index type, and all numbers other than 1 and -1. 1 In 2.4, creating any invalid index will result in an error. Furthermore, you cannot create a 2dsphere or text index on a collection if its containing database has any invalid index types. 1 Example If you attempt to add an invalid index in MongoDB 2.4, as in the following:
db.coll.ensureIndex( { field: "1" } )
MongoDB will return the following error document:

{ "err" : "Unknown index plugin 1 in index { field: \"1\" }" "code": 16734, "n": <number>, "connectionId": <number>, "ok": 1 }
See Upgrade MongoDB to 2.4 (page 1137) for full upgrade instructions.
67.1.4 Changes
Major Features
Text Search
MongoDB 2.4 adds support for boolean search of content in MongoDB databases as a beta feature. With the new text index (page 338), and supporting, text (page 832) command you can search text in data stored in MongoDB, using
In 2.4, indexes that specify a type of "1" or "-1" (the strings "1" and "-1") will continue to exist, despite a warning on start-up. However, a secondary in a replica set cannot complete an initial sync from a primary that has a "1" or "-1" index. Avoid all indexes with invalid types.
1
1144
an index that updates in real-time and is always consistent with the data set. See Text Search (page 371) for more information about text search in MongoDB.
New Geospatial Indexes with GeoJSON and Improved Spherical Geometry
MongoDB adds the new 2dsphere geospatial index in addition to the existing 2d index. The 2dsphere index supports improved spherical queries and supports the following GeoJSON objects: Point LineString Polygon The 2dsphere index supports all current geospatial query operators (page 777) and introduces the following new query operator for queries on GeoJSON data: $geoWithin (page 778) operator $geoIntersects (page 777) operator The operators use the new $geometry parameter. The $within (page 779) operator no longer requires a geospatial index. Additionally, 2.4 deprecates the $within (page 779) operator. Use $geoWithin (page 778) operator instead. For more information on geospatial indexes in 2.4, see: Geospatial Indexes and Queries (page 357) 2d Index Internals (page 369)
New Hashed Index and Sharding with a Hashed Shard Key
To support an easy to congure and evenly distributed shard key, version 2.4 adds a new hashed index type that indexes documents using hashes of eld values. See Hashed Index (page 335) for documentation of hashed indexes, and Hashed Sharding (page 486) for documentation of hash-based sharding. Security Improvements
New Modular Authentication System with Support for Kerberos
Note: Kerberos authentication is only present in MongoDB Enterprise Edition. To download and install MongoDB Enterprise, see Install MongoDB Enterprise (page 20). In 2.4 the MongoDB Enterprise now supports authentication via a Kerberos mechanism. See Deploy MongoDB with Kerberos Authentication (page 149) for more information. Also consider the following documents that address authenticating to MongoDB using Kerberos: Authenticate to MongoDB using Kerberos and the Java Driver Authenticate to MongoDB using Kerberos and the C# Driver See also: MongoDB Security Practices and Procedures (page 129). 67.1. Release Notes for MongoDB 2.4 1145
Role Based Access Control and New Privilege Documents
MongoDB 2.4 introduces a role based access control system that provides more granular privileges to MongoDB users. See User Privilege Roles in MongoDB (page 155) for more information. To support the new access control system, 2.4 also introduces a new format for documents in a databases system.users (page 160) collection. See system.users Privilege Documents (page 159) for more information. Use supportCompatibilityFormPrivilegeDocuments (page 1094) to disable the legacy privilege documents, which MongoDB continues to support in 2.4.
Enhanced SSL Support
In 2.4, MongoDB instances can optionally require clients to provide SSL certicates signed by a Certicate Authority. You must use the MongoDB distribution that supports SSL, and your client driver must support SSL. See Connect to MongoDB with SSL (page 75) for more information.
Compatibility Change: User Uniqueness Enforced
2.4 now enforces uniqueness of the user eld in user privilege documents (i.e. in the system.users (page 160) collection.) Previous versions of MongoDB did not enforce this requirement, and existing databases may have duplicates. Administration Changes
--setParameter Option Available on the mongos and mongod Command Line
You can now use --setParameter (page 1027) on the command line and setParameter (page 1087) in the conguration le. For mongod (page 1021) the following options are available using setParameter (page 1087): enableLocalhostAuthBypass (page 1093) enableTestCommands (page 1093) journalCommitInterval (page 1093) logLevel (page 1094) logUserIds (page 1094) notablescan (page 1094) quiet (page 1095) replApplyBatchSize (page 1094) replIndexPrefetch (page 1094) supportCompatibilityFormPrivilegeDocuments (page 1094) syncdelay (page 1094) textSearchEnabled (page 1095) traceExceptions (page 1095) For mongos (page 1032) the following options are available using setParameter (page 1087): enableLocalhostAuthBypass (page 1093)
1146
enableTestCommands (page 1093) logLevel (page 1094) logUserIds (page 1094) notablescan (page 1094) quiet (page 1095) supportCompatibilityFormPrivilegeDocuments (page 1094) syncdelay (page 1094) textSearchEnabled (page 1095) See mongod Parameters (page 1092) for full documentation of available parameters and their use.
Changes to serverStatus Output Including Additional Metrics
In 2.4 MongoDB adds a number of counters and system metrics to the output of the serverStatus (page 889) command, including: a working set estimator (page 902). operation counters, in document (page 903) and operation (page 904). record allocation, in record (page 904). thorough metrics of the replication process, in repl (page 904). metrics on the ttl index (page 577) documentation. Additionally, in 2.4, the serverStatus (page 889) command can dynamically construct the serverStatus (page 889) document by excluding any top-level sections included by default, or including any top-level section not included by default (e.g. workingSet (page 902).) See db.serverStatus() (page 984) and serverStatus (page 889) for more information.
Increased Chunk Migration Write Concern
By default, all insert and delete operations that occur as part of a chunk migration in a sharded cluster will have an increased write concern, to ensure that at least one secondary acknowledges each insert and deletion operation. This change slows the potential speed of a chunk migration, but increases reliability and ensures that a large number of chunk migrations cannot affect the availability of a sharded cluster.
BSON Document Validation Enabled by Default for mongod and mongorestore
Starting in 2.4, MongoDB enables basic BSON object validation for mongod (page 1021) and mongorestore (page 1048) when writing to MongoDB data les. This prevents any client from inserting invalid or malformed BSON into a MongoDB database. For objects with a high degree of sub-document nesting this validation may have a small performance impact. objcheck (page 1080), which was previously disabled by default, provides this validation.
1147
Indexing Changes
Support for Multiple Concurrent Index Builds
A single mongod (page 1021) instance can build multiple indexes in the background at the same time. See building indexes in the background (page 336) for more information on background index builds. Foreground index builds hold a database lock and must proceed one at a time.
db.killOp() Can Now Kill Foreground Index Builds
The db.killOp() (page 981) method will now terminate a foreground index build, in addition to the other operations supported in previous versions.
Improved Validation of Index Types
Before 2.4, mongod (page 1021) would create an ascending scalar index (e.g. { a : 1 }) when users attempted to create an index of a type that did not exist. Creating an index of an invalid index type will generate an error in 2.4. See Compatibility and Index Type Changes in MongoDB 2.4 (page 1144) for more information. Interface Changes
$setOnInsert New Update Operator
To set elds only when an upsert (page 948) performs an insert, use the $setOnInsert (page 787) operator with the upsert (page 948) . Example A collection named coll has no documents with _id equal to 1. The following upsert (page 948) operation inserts a document and applies the $setOnInsert (page 787) operator to set the elds x and y:
db.coll.update( { _id: 1 }, { $setOnInsert: { x: 25, y: 30 } }, { upsert: true } )
The newly-inserted document has the eld x set to 25 and the eld y set to 30:
{ "_id" : 1, "x" : 25, "y" : 30 }
Note: The $setOnInsert (page 787) operator performs no operation for upserts (page 948) that only perform an update and for updates (page 948) when the upsert option is false.
Limit Number of Elements in an Array
In 2.4, by using the $push (page 792) operator with the $each (page 793), the $sort (page 794), and the $slice (page 793) modiers, you can add multiple elements to an array, sort and limit the number of elements in the modied array to maintain an array with a xed number of elements.
1148
See Limit Number of Elements in an Array after an Update (page 575) for an example where an update maintains the top three scores for a student. See also: The following pages provide additional information and examples: $push (page 792) operator $each (page 793) modier $sort (page 794) modier $slice (page 793) modier
JavaScript Engine Changed to V8
JavaScript Changes in MongoDB 2.4 (page 1149) in MongoDB 2.4:
Consider the following impacts of JavaScript Engine Changed to V8
Improved Concurrency Previously, MongoDB operations that required the JavaScript interpreter had to acquire a lock, and a single mongod (page 1021) could only run a single JavaScript operation at a time. The switch to V8 improves concurrency by permitting multiple JavaScript operations to run at the same time. Modernized JavaScript Implementation (ES5) The 5th edition of ECMAscript, abbreviated as ES5, adds many new language features, including: standardized JSON, strict mode, function.bind(), array extensions, and getters and setters. With V8, MongoDB supports the ES5 implementation of Javascript with the following exceptions. Note: The following features do not work as expected on documents returned from MongoDB queries: Object.seal() throws an exception on documents returned from MongoDB queries. Object.freeze() throws an exception on documents returned from MongoDB queries. Object.preventExtensions() incorrectly allows the addition of new properties on documents returned from MongoDB queries. enumerable properties, when added to documents returned from MongoDB queries, are not saved during write operations. See SERVER-8216, SERVER-8223, SERVER-8215, and SERVER-8214 for more information. For objects that have not been returned from MongoDB queries, the features work as expected.
Removed Non-Standard SpiderMonkey Features V8 does not support the following non-standard SpiderMonkey JavaScript extensions, previously supported by MongoDBs use of SpiderMonkey as its JavaScript engine.
1149
E4X Extensions V8 does not support the non-standard E4X extensions. E4X provides a native XML object to the JavaScript language and adds the syntax for embedding literal XML documents in JavaScript code. You need to use alternative XML processing if you used any of the following constructors/methods: XML() Namespace() QName() XMLList() isXMLName() Destructuring Assignment V8 does not support the non-standard destructuring assignments. Destructuring assignment extract[s] data from arrays or objects using a syntax that mirrors the construction of array and object literals. Mozilla docs Example The following destructuring assignment is invalid with V8 and throws a SyntaxError:
original = [4, 8, 15]; var [b, ,c] = a; // <== destructuring assignment print(b) // 4 print(c) // 15
Iterator(), StopIteration(), and Generators V8 does not support Iterator(), StopIteration(), and generators. InternalError() V8 does not support InternalError(). Use Error() instead. for each...in Construct V8 does not support the use of for each...in construct. Use for (var x in y) construct instead. Example The following for each (var x in y) construct is invalid with V8:
var o = { name: MongoDB, version: 2.4 }; for each (var value in o) { print(value); }
Instead, in version 2.4, you can use the for (var x in y) construct:
var o = { name: MongoDB, version: 2.4 }; for (var prop in o) { var value = o[prop]; print(value); }
You can also use the array instance method forEach() with the ES5 method Object.keys():
1150
Object.keys(o).forEach(function (key) { var value = o[key]; print(value); });
Array Comprehension V8 does not support Array comprehensions. Use other methods such as the Array instance methods map(), filter(), or forEach(). Example With V8, the following array comprehension is invalid:
var a = { w: 1, x: 2, y: 3, z: 4 } var arr = [i * i for each (i in a) if (i > 2)] printjson(arr)
Instead, you can implement using the Array instance method forEach() and the ES5 method Object.keys() :
var a = { w: 1, x: 2, y: 3, z: 4 } var arr = []; Object.keys(a).forEach(function (key) { var val = a[key]; if (val > 2) arr.push(val * val); }) printjson(arr)
Note: The new logic uses the Array instance method forEach() and not the generic method Array.forEach(); V8 does not support Array generic methods. See Array Generic Methods (page 1153) for more information.
Multiple Catch Blocks V8 does not support multiple catch blocks and will throw a SyntaxError. Example The following multiple catch blocks is invalid with V8 and will throw "SyntaxError: if":
try { something() } catch (err if err instanceof SomeError) { print(some error) } catch (err) { print(standard error) }
Unexpected token
Conditional Function Denition V8 will produce different outcomes than SpiderMonkey with conditional function denitions. Example 67.1. Release Notes for MongoDB 2.4 1151
The following conditional function denition produces different outcomes in SpiderMonkey versus V8:
function test () { if (false) { function go () {}; } print(typeof go) }
With SpiderMonkey, the conditional function outputs undefined, whereas with V8, the conditional function outputs function. If your code denes functions this way, it is highly recommended that you refactor the code. The following example refactors the conditional function denition to work in both SpiderMonkey and V8.
function test () { var go; if (false) { go = function () {} } print(typeof go) }
The refactored code outputs undefined in both SpiderMonkey and V8. Note: ECMAscript prohibits conditional function denitions. To force V8 to throw an Error, enable strict mode.
function test () { use strict; if (false) { function go () {} } }
The JavaScript code throws the following syntax error:
SyntaxError: In strict mode code, functions can only be declared at top level or immediately within a
String Generic Methods V8 does not support String generics. String generics are a set of methods on the String class that mirror instance methods. Example The following use of the generic method String.toLowerCase() is invalid with V8:
var name = MongoDB; var lower = String.toLowerCase(name);
With V8, use the String instance method toLowerCase() available through an instance of the String class instead:
var name = MongoDB; var lower = name.toLowerCase(); print(name + becomes + lower);
1152
With V8, use the String instance methods instead of following generic methods: String.charAt() String.charCodeAt() String.concat() String.endsWith() String.indexOf() String.lastIndexOf() String.localeCompare() String.match() String.quote() String.replace() String.search() String.slice() String.split() String.startsWith() String.substr() String.substring() String.toLocaleLowerCase() String.toLocaleUpperCase() String.toLowerCase() String.toUpperCase() String.trim() String.trimLeft() String.trimRight()
Array Generic Methods V8 does not support Array generic methods. Array generics are a set of methods on the Array class that mirror instance methods. Example The following use of the generic method Array.every() is invalid with V8:
var arr = [4, 8, 15, 16, 23, 42]; function isEven (val) { return 0 === val % 2; } var allEven = Array.every(arr, isEven); print(allEven);
With V8, use the Array instance method every() available through an instance of the Array class instead:
var allEven = arr.every(isEven); print(allEven);
With V8, use the Array instance methods instead of the following generic methods: Array.concat() Array.every() Array.filter() Array.forEach() Array.indexOf() Array.join() Array.lastIndexOf() Array.map() Array.pop() Array.push() Array.reverse() Array.shift() Array.slice() Array.some() Array.sort() Array.splice() Array.unshift()
Array Instance Method toSource() V8 does not support the Array instance method toSource(). Use the Array instance method toString() instead. uneval() V8 does not support the non-standard method uneval(). Use the standardized JSON.stringify() method instead. In 2.4 the default JavaScript engine in the mongo (page 1036) shell mongod (page 1021) is now V8. This change affects all JavaScript behavior including the mapReduce (page 814), group (page 810), and eval (page 823) commands, as well as the $where (page 775) query operator. Use the new interpreterVersion() method in the mongo (page 1036) shell and the javascriptEngine (page 872) eld in the output of db.serverBuildInfo() (page 984) to determine which JavaScript engine a MongoDB binary uses.
1153
The primary impacts of the change from the previous JavaScript engine, SpiderMonkey, to V8 are: improved concurrency for JavaScript operations, modernized JavaScript implementation, and removed non-standard SpiderMonkey features. See JavaScript Changes in MongoDB 2.4 (page 1149) for more information about all changes .
Additional Limitations for Map-Reduce and $where Operations
In MongoDB 2.4, map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions cannot access certain global functions or properties, such as db, that are available in the mongo (page 1036) shell. When upgrading to MongoDB 2.4, you will need to refactor your code if your map-reduce operations (page 814), group (page 810) commands, or $where (page 775) operator expressions include any global shell functions or properties that are no longer available, such as db. The following shell functions and properties are available to map-reduce operations (page 814), the group (page 810) command, and $where (page 775) operator expressions in MongoDB 2.4: Available Properties args MaxKey MinKey Available Functions assert() BinData() DBPointer() DBRef() doassert() emit() gc() HexData() hex_md5() isNumber() isObject() ISODate() isString() Map() MD5() NumberInt() NumberLong() ObjectId() print() printjson() printjsononeline() sleep() Timestamp() tojson() tojsononeline() tojsonObject() UUID() version()
Improvements to the Aggregation Framework
MongoDB 2.4 introduces a number of additional functionality and improved performance for the Aggregation Framework (page 255). Consider the following additions in 2.4: $match (page 281) queries now support the $geoWithin (page 778) operator for bounded geospatial queries. The new $geoNear (page 276) pipeline stage to support geospatial queries. $min (page 283) operator only considers non-null and existing eld values. If all the values for a eld are null or are missing, the operator returns null for the minimum value.
1154
For sort operations where the $sort (page 287) stage immediately precedes a $limit (page 280) in the pipeline, the MongoDB can perform a more efcient sort that does not require keeping the entire result set in memory. The new $millisecond (page 282) operator returns the millisecond portion of a date. The new $concat (page 272) operator concatenates array of strings.

MongoDB Downloads. Whats New in MongoDB 2.4 (page 1135). All JIRA issues resolved in 2.4. All Backwards incompatible changes. All Third Party License Notices. See http://docs.mongodb.org/manual/release-notes/2.4-changes for an overview of all changes in 2.4.
1155
1156
CHAPTER 68
Previous Stable Releases

See the full index of this page for a complete list of changes included in 2.2. Upgrading (page 1157) Changes (page 1159) Licensing Changes (page 1166) Resources (page 1166)
68.1.1 Upgrading
MongoDB 2.2 is a production release series and succeeds the 2.0 production release series. MongoDB 2.0 data les are compatible with 2.2-series binaries without any special migration process. However, always perform the upgrade process for replica sets and sharded clusters using the procedures that follow. Always upgrade to the latest point release in the 2.2 point release. Currently the latest release of MongoDB is 2.4.3. Synopsis mongod (page 1021), 2.2 is a drop-in replacement for 2.0 and 1.8. Check your driver (page 555) documentation for information regarding required compatibility upgrades, and always run the recent release of your driver. Typically, only users running with authentication, will need to upgrade drivers before continuing with the upgrade to 2.2. For all deployments using authentication, upgrade the drivers (i.e. client libraries), before upgrading the mongod (page 1021) instance or instances. For all upgrades of sharded clusters: turn off the balancer during the upgrade process. See the Disable the Balancer (page 529) section for more information. upgrade all mongos (page 1032) instances before upgrading any mongod (page 1021) instances.
1157
Other than the above restrictions, 2.2 processes can interoperate with 2.0 and 1.8 tools and processes. You can safely upgrade the mongod (page 1021) and mongos (page 1032) components of a deployment one by one while the deployment is otherwise operational. Be sure to read the detailed upgrade procedures below before upgrading production systems. Upgrading a Standalone mongod 1. Download binaries of the latest release in the 2.2 series from the MongoDB Download Page. 2. Shutdown your mongod (page 1021) instance. Replace the existing binary with the 2.2 mongod (page 1021) binary and restart MongoDB. Upgrading a Replica Set You can upgrade to 2.2 by performing a rolling upgrade of the set by upgrading the members individually while the other members are available to minimize downtime. Use the following procedure: 1. Upgrade the secondary members of the set one at a time by shutting down the mongod (page 1021) and replacing the 2.0 binary with the 2.2 binary. After upgrading a mongod (page 1021) instance, wait for the member to recover to SECONDARY state before upgrading the next instance. To check the members state, issue rs.status() (page 990) in the mongo (page 1036) shell. 2. Use the mongo (page 1036) shell method rs.stepDown() (page 991) to step down the primary to allow the normal failover (page 389) procedure. rs.stepDown() (page 991) expedites the failover procedure and is preferable to shutting down the primary directly. Once the primary has stepped down and another member has assumed PRIMARY state, as observed in the output of rs.status() (page 990), shut down the previous primary and replace mongod (page 1021) binary with the 2.2 binary and start the new process. Note: Replica set failover is not instant but will render the set unavailable to read or accept writes until the failover process completes. Typically this takes 10 seconds or more. You may wish to plan the upgrade during a predened maintenance window.
Upgrading a Sharded Cluster Use the following procedure to upgrade a sharded cluster: Disable the balancer (page 529). Upgrade all mongos (page 1032) instances rst, in any order. Upgrade all of the mongod (page 1021) cong server instances using the stand alone (page 1158) procedure. To keep the cluster online, be sure that at all times at least one cong server is up. Upgrade each shards replica set, using the upgrade procedure for replica sets (page 1158) detailed above. re-enable the balancer. Note: Balancing is not currently supported in mixed 2.0.x and 2.2.0 deployments. Thus you will want to reach a consistent version for all shards within a reasonable period of time, e.g. same-day. See SERVER-6902 for more information.
1158
Chapter 68. Previous Stable Releases
68.1.2 Changes
Major Features
Aggregation Framework
The aggregation framework makes it possible to do aggregation operations without needing to use map-reduce. The aggregate (page 808) command exposes the aggregation framework, and the db.collection.aggregate() (page 918) helper in the mongo (page 1036) shell provides an interface to these operations. Consider the following resources for background on the aggregation framework and its use: Documentation: Aggregation Framework (page 255) Reference: Aggregation Framework Reference (page 271) Examples: Aggregation Framework Examples (page 261)
TTL Collections
TTL collections remove expired data from a collection, using a special index and a background thread that deletes expired documents every minute. These collections are useful as an alternative to capped collections in some cases, such as for data warehousing and caching cases, including: machine generated event data, logs, and session information that needs to persist in a database for only a limited period of time. For more information, see the Expire Data from Collections by Setting TTL (page 577) tutorial.
Concurrency Improvements
MongoDB 2.2 increases the servers capacity for concurrent operations with the following improvements: 1. DB Level Locking 2. Improved Yielding on Page Faults 3. Improved Page Fault Detection on Windows To reect these changes, MongoDB now provides changed and improved reporting for concurrency and use, see locks (page 890) and recordStats (page 901) in server status (page 889) and see db.currentOp() (page 971), mongotop (page 1067), and mongostat (page 1063).
Improved Data Center Awareness with Tag Aware Sharding
MongoDB 2.2 adds additional support for geographic distribution or other custom partitioning for sharded collections in clusters. By using this tag aware sharding, you can automatically ensure that data in a sharded database system is always on specic shards. For example, with tag aware sharding, you can ensure that data is closest to the application servers that use that data most frequently. Shard tagging controls data location, and is complementary but separate from replica set tagging, which controls read preference (page 402) and write concern (page 398). For example, shard tagging can pin all USA data to one or more logical shards, while replica set tagging can control which mongod (page 1021) instances (e.g. production or reporting) the application uses to service requests. See the documentation for the following helpers in the mongo (page 1036) shell that support tagged sharding conguration: sh.addShardTag() (page 996)
1159
sh.addTagRange() sh.removeShardTag() (page 1000) Also, see Tag Aware Sharding (page 532) and Administer and Manage Shard Tags (page 517).
Fully Supported Read Preference Semantics
All MongoDB clients and drivers now support full read preferences (page 402), including consistent support for a full range of read preference modes (page 402) and tag sets (page 404). This support extends to the mongos (page 1032) and applies identically to single replica sets and to the replica sets for each shard in a sharded cluster. Additional read preference support now exists in the mongo (page 1036) shell using the readPref() (page 963) cursor method. Compatibility Changes
Authentication Changes
MongoDB 2.2 provides more reliable and robust support for authentication clients, including drivers and mongos (page 1032) instances. If your cluster runs with authentication: For all drivers, use the latest release of your driver and check its release notes. In sharded environments, to ensure that your cluster remains available during the upgrade process you must use the upgrade procedure for sharded clusters (page 1158).
findAndModify Returns Null Value for Upserts that Perform Inserts
In version 2.2, for upsert that perform inserts with the new option set to false, findAndModify (page 826) commands will now return the following output:
{ ok: 1.0, value: null }
In the mongo (page 1036) shell, upsert findAndModify (page 826) operations that perform inserts (with new set to false.)only output a null value. In version 2.0 these operations would return an empty document, e.g. { }. See: SERVER-6226 for more information.
mongodump Output can only Restore to 2.2 MongoDB Instances
If you use the mongodump (page 1044) tool from the 2.2 distribution to create a dump of a database, you must use a 2.2 version of mongorestore (page 1048) to restore that dump. See: SERVER-6961 for more information.
1160
ObjectId().toString() Returns String Literal ObjectId("...")
In version 2.2, the ObjectId.toString() method returns the string representation of the ObjectId() (page 194) object and has the format ObjectId("..."). Consider the following example that calls ObjectId("507c7f79bcf86cd7994f6c0e") object:
ObjectId("507c7f79bcf86cd7994f6c0e").toString()
the
toString()
method
on
the
The method now returns the string ObjectId("507c7f79bcf86cd7994f6c0e"). Previously, in version 2.0, the method would return the hexadecimal string 507c7f79bcf86cd7994f6c0e. If compatibility between versions 2.0 and 2.2 is required, use ObjectId().str (page 194), which holds the hexadecimal string value in both versions.
ObjectId().valueOf() Returns hexadecimal string
In version 2.2, the ObjectId.valueOf() method returns the value of the ObjectId() (page 194) object as a lowercase hexadecimal string. Consider the following example that calls ObjectId("507c7f79bcf86cd7994f6c0e") object:
ObjectId("507c7f79bcf86cd7994f6c0e").valueOf()
the
valueOf()
method
on
the
The method now returns the hexadecimal string 507c7f79bcf86cd7994f6c0e. Previously, in version 2.0, the method would return the object ObjectId("507c7f79bcf86cd7994f6c0e"). If compatibility between versions 2.0 and 2.2 is required, use ObjectId().str (page 194) attribute, which holds the hexadecimal string value in both versions. Behavioral Changes
Restrictions on Collection Names
In version 2.2, collection names cannot: contain the $. be an empty string (i.e. ""). This change does not affect collections created with now illegal names in earlier versions of MongoDB. These new restrictions are in addition to the existing restrictions on collection names which are: A collection name should begin with a letter or an underscore. A collection name cannot contain the null character. Begin with the system. prex. system.indexes collection. MongoDB reserves system. for system collections, such as the
The maximum size of a collection name is 128 characters, including the name of the database. However, for maximum exibility, collections should have names less than 80 characters. Collections names may have any other valid UTF-8 string. See the SERVER-4442 and the Are there any restrictions on the names of Collections? (page 720) FAQ item. 68.1. Release Notes for MongoDB 2.2 1161
Restrictions on Database Names for Windows
Database names running on Windows can no longer contain the following characters:
/\. "*<>:|?
The names of the data les include the database name. If you attempt to upgrade a database instance with one or more of these characters, mongod (page 1021) will refuse to start. Change the name of these databases before upgrading. See SERVER-4584 and SERVER-6729 for more information.
_id Fields and Indexes on Capped Collections
All capped collections now have an _id eld by default, if they exist outside of the local database, and now have indexes on the _id eld. This change only affects capped collections created with 2.2 instances and does not affect existing capped collections. See: SERVER-5516 for more information.
New $elemMatch Projection Operator
The $elemMatch (page 798) operator allows applications to narrow the data returned from queries so that the query operation will only return the rst matching element in an array. See the $elemMatch (projection) (page 798) documentation and the SERVER-2238 and SERVER-828 issues for more information. Windows Specic Changes
Windows XP is Not Supported
As of 2.2, MongoDB does not support Windows XP. Please upgrade to a more recent version of Windows to use the latest releases of MongoDB. See SERVER-5648 for more information.
Service Support for mongos.exe
You may now run mongos.exe (page 1042) instances as a Windows Service. See the mongos.exe (page 1042) reference and MongoDB as a Windows Service (page 18) and SERVER-1589 for more information.
Log Rotate Command Support
MongoDB for Windows now supports log rotation by way of the logRotate (page 866) database command. See SERVER-2612 for more information.
New Build Using SlimReadWrite Locks for Windows Concurrency
Labeled 2008+ on the Downloads Page, this build for 64-bit versions of Windows Server 2008 R2 and for Windows 7 or newer, offers increased performance over the standard 64-bit Windows build of MongoDB. See SERVER-3844 for more information.
1162
Tool Improvements
Index Denitions Handled by mongodump and mongorestore
When you specify the --collection (page 1046) option to mongodump (page 1044), mongodump (page 1044) will now backup the denitions for all indexes that exist on the source database. When you attempt to restore this backup with mongorestore (page 1048), the target mongod (page 1021) will rebuild all indexes. See SERVER808 for more information. mongorestore (page 1048) now includes the --noIndexRestore (page 1050) option to provide the preceding behavior. Use --noIndexRestore (page 1050) to prevent mongorestore (page 1048) from building previous indexes.
mongooplog for Replaying Oplogs
The mongooplog (page 1053) tool makes it possible to pull oplog entries from mongod (page 1021) instance and apply them to another mongod (page 1021) instance. You can use mongooplog (page 1053) to achieve point-in-time backup of a MongoDB data set. See the SERVER-3873 case and the mongooplog (page 1052) documentation.
Authentication Support for mongotop and mongostat
mongotop (page 1068) and mongostat (page 1063) now contain support for username/password authentication. See SERVER-3875 and SERVER-3871 for more information regarding this change. Also consider the documentation of the following options for additional information: mongotop --username (page 1068) mongotop --password (page 1068) mongostat --username (page 1064) mongostat --password (page 1064)
Write Concern Support for mongoimport and mongorestore
mongoimport (page 1056) now provides an option to halt the import if the operation encounters an error, such as a network interruption, a duplicate key exception, or a write error. The --stopOnError (page 1058) option will produce an error rather than silently continue importing data. See SERVER-3937 for more information. In mongorestore (page 1048), the --w (page 1050) option provides support for congurable write concern.
mongodump Support for Reading from Secondaries
You can now run mongodump (page 1044) when connected to a secondary member of a replica set. See SERVER3854 for more information.
mongoimport Support for full 16MB Documents
Previously, mongoimport (page 1056) would only import documents that were less than 4 megabytes in size. This issue is now corrected, and you may use mongoimport (page 1056) to import documents that are at least 16 megabytes ins size. See SERVER-4593 for more information.
1163
Timestamp() Extended JSON format
MongoDB extended JSON now includes a new Timestamp() type to represent the Timestamp type that MongoDB uses for timestamps in the oplog among other contexts. This permits tools like mongooplog (page 1053) and mongodump (page 1044) to query for specic timestamps. Consider the following mongodump (page 1044) operation:
mongodump --db local --collection oplog.rs --query {"ts":{"$gt":{"$timestamp" : {"t": 1344969612000,
See SERVER-3483 for more information. Shell Improvements

Improved Shell User Interface
2.2 includes a number of changes that improve the overall quality and consistency of the user interface for the mongo (page 1036) shell: Full Unicode support. Bash-like line editing features. See SERVER-4312 for more information. Multi-line command support in shell history. See SERVER-3470 for more information. Windows support for the edit command. See SERVER-3998 for more information.
Helper to load Server-Side Functions
The db.loadServerScripts() (page 982) loads the contents of the current databases system.js collection into the current mongo (page 1036) shell session. See SERVER-1651 for more information.
Support for Bulk Inserts
If you pass an array of documents to the insert() (page 936) method, the mongo (page 1036) shell will now perform a bulk insert operation. See SERVER-3819 and SERVER-2395 for more information. Note: For bulk inserts on sharded clusters, the getLastError (page 831) command alone is insufcient to verify success. Applications should must verify the success of bulk inserts in application logic.
Operations
Support for Logging to Syslog
See the SERVER-2957 case and the documentation of the syslog (page 1080) run-time option or the mongod --syslog (page 1022) and mongos --syslog (page 1033) command line-options.
touch Command
Added the touch (page 870) command to read the data and/or indexes from a collection into memory. See: SERVER2023 and touch (page 870) for more information.
1164
indexCounters No Longer Report Sampled Data
indexCounters now report actual counters that reect index use and state. In previous versions, these data were sampled. See SERVER-5784 and indexCounters for more information.
Padding Speciable on compact Command
See the documentation of the compact (page 859) and the SERVER-4018 issue for more information.
Added Build Flag to Use System Libraries
The Boost library, version 1.49, is now embedded in the MongoDB code base. If you want to build MongoDB binaries using system Boost libraries, you can pass scons using the --use-system-boost ag, as follows:
scons --use-system-boost
When building MongoDB, you can also pass scons a ag to compile MongoDB using only system libraries rather than the included versions of the libraries. For example:
scons --use-system-all
See the SERVER-3829 and SERVER-5172 issues for more information.

Memory Allocator Changed to TCMalloc
To improve performance, MongoDB 2.2 uses the TCMalloc memory allocator from Google Perftools. For more information about this change see the SERVER-188 and SERVER-4683. For more information about TCMalloc, see the documentation of TCMalloc itself. Replication
Improved Logging for Replica Set Lag
When secondary members of a replica set fall behind in replication, mongod (page 1021) now provides better reporting in the log. This makes it possible to track replication in general and identify what process may produce errors or halt replication. See SERVER-3575 for more information.
Replica Set Members can Sync from Specic Members
The new replSetSyncFrom (page 845) command and new rs.syncFrom() (page 991) helper in the mongo (page 1036) shell make it possible for you to manually congure from which member of the set a replica will poll oplog entries. Use these commands to override the default selection logic if needed. Always exercise caution with replSetSyncFrom (page 845) when overriding the default behavior.
1165
Replica Set Members will not Sync from Members Without Indexes Unless buildIndexes:
false
To prevent inconsistency between members of replica sets, if the member of a replica set has buildIndexes (page 464) set to true, other members of the replica set will not sync from this member, unless they also have buildIndexes (page 464) set to true. See SERVER-4160 for more information.
New Option To Congure Index Pre-Fetching during Replication
By default, when replicating options, secondaries will pre-fetch Indexes (page 327) associated with a query to improve replication throughput in most cases. The replIndexPrefetch (page 1088) setting and --replIndexPrefetch (page 1029) option allow administrators to disable this feature or allow the mongod (page 1021) to pre-fetch only the index on the _id eld. See SERVER-6718 for more information. Map Reduce Improvements In 2.2 Map Reduce received the following improvements: Improved support for sharded MapReduce, and MapReduce will retry jobs following a cong error. Sharding Improvements
Index on Shard Keys Can Now Be a Compound Index
If your shard key uses the prex of an existing index, then you do not need to maintain a separate index for your shard key in addition to your existing index. This index, however, cannot be a multi-key index. See the Shard Key Indexes (page 496) documentation and SERVER-1506 for more information.
Migration Thresholds Modied
The migration thresholds (page 497) have changed in 2.2 to permit more even distribution of chunks in collections that have smaller quantities of data. See the Migration Thresholds (page 497) documentation for more information.
68.1.3 Licensing Changes

Added License notice for Google Perftools (TCMalloc Utility). See the License Notice and the SERVER-4683 for more information.
68.1.4 Resources
MongoDB Downloads. All JIRA issues resolved in 2.2. All backwards incompatible changes. All third party license notices. Whats New in MongoDB 2.2 Online Conference.
1166

See the full index of this page for a complete list of changes included in 2.0. Upgrading (page 1167) Changes (page 1168) Resources (page 1173)
68.2.1 Upgrading
Although the major version number has changed, MongoDB 2.0 is a standard, incremental production release and works as a drop-in replacement for MongoDB 1.8. Preparation Read through all release notes before upgrading, and ensure that no changes will affect your deployment. If you create new indexes in 2.0, then downgrading to 1.8 is possible but you must reindex the new collections. mongoimport (page 1056) and mongoexport (page 1059) now correctly adhere to the CSV spec for handling CSV input/output. This may break existing import/export workows that relied on the previous behavior. For more information see SERVER-1097. Journaling is enabled by default in 2.0 for 64-bit builds. If you still prefer to run without journaling, start mongod (page 1021) with the --nojournal (page 1026) run-time option. Otherwise, MongoDB creates journal les during startup. The rst time you start mongod (page 1021) with journaling, you will see a delay as mongod (page 1021) creates new les. In addition, you may see reduced write throughput. 2.0 mongod (page 1021) instances are interoperable with 1.8 mongod (page 1021) instances; however, for best results, upgrade your deployments using the following procedures: Upgrading a Standalone mongod 1. Download the v2.0.x binaries from the MongoDB Download Page. 2. Shutdown your mongod (page 1021) instance. Replace the existing binary with the 2.0.x mongod (page 1021) binary and restart MongoDB. Upgrading a Replica Set 1. Upgrade the secondary members of the set one at a time by shutting down the mongod (page 1021) and replacing the 1.8 binary with the 2.0.x binary from the MongoDB Download Page. 2. To avoid losing the last few updates on failover you can temporarily halt your application (failover should take less than 10 seconds), or you can set write concern (page 398) in your application code to conrm that each update reaches multiple servers. 3. Use the rs.stepDown() (page 991) to step down the primary to allow the normal failover (page 389) procedure. rs.stepDown() (page 991) and replSetStepDown (page 844) provide for shorter and more consistent failover procedures than simply shutting down the primary directly.
1167
When the primary has stepped down, shut down its instance and upgrade by replacing the mongod (page 1021) binary with the 2.0.x binary. Upgrading a Sharded Cluster 1. Upgrade all cong server instances rst, in any order. Since cong servers use two-phase commit, shard conguration metadata updates will halt until all are up and running. 2. Upgrade mongos (page 1032) routers in any order.
68.2.2 Changes
Compact Command A compact (page 859) command is now available for compacting a single collection and its indexes. Previously, the only way to compact was to repair the entire database. Concurrency Improvements When going to disk, the server will yield the write lock when writing data that is not likely to be in memory. The initial implementation of this feature now exists: See SERVER-2563 for more information. The specic operations yield in 2.0 are: Updates by _id Removes Long cursor iterations Default Stack Size MongoDB 2.0 reduces the default stack size. This change can reduce total memory usage when there are many (e.g., 1000+) client connections, as there is a thread per connection. While portions of a threads stack can be swapped out if unused, some operating systems do this slowly enough that it might be an issue. The default stack size is lesser of the system setting or 1MB. Index Performance Enhancements v2.0 includes signicant improvements to the index (page 356). Indexes are often 25% smaller and 25% faster (depends on the use case). When upgrading from previous versions, the benets of the new index type are realized only if you create a new index or re-index an old one. Dates are now signed, and the max index key size has increased slightly from 819 to 1024 bytes. All operations that create a new index will result in a 2.0 index by default. For example: Reindexing results on an older-version index results in a 2.0 index. However, reindexing on a secondary does not work in versions prior to 2.0. Do not reindex on a secondary. For a workaround, see SERVER-3866. The repairDatabase command converts indexes to a 2.0 indexes.
1168
To convert all indexes for a given collection to the 2.0 type (page 1168), invoke the compact (page 859) command. Once you create new indexes, downgrading to 1.8.x will require a re-index of any indexes created using 2.0. See Build Old Style Indexes (page 356). Sharding Authentication Applications can now use authentication with sharded clusters. Replica Sets
Hidden Nodes in Sharded Clusters
In 2.0, mongos (page 1032) instances can now determine when a member of a replica set becomes hidden without requiring a restart. In 1.8, mongos (page 1032) if you recongured a member as hidden, you had to restart mongos (page 1032) to prevent queries from reaching the hidden member.
Priorities
Each replica set member can now have a priority value consisting of a oating-point from 0 to 1000, inclusive. Priorities let you control which member of the set you prefer to have as primary the member with the highest priority that can see a majority of the set will be elected primary. For example, suppose you have a replica set with three members, A, B, and C, and suppose that their priorities are set as follows: As priority is 2. Bs priority is 3. Cs priority is 1. During normal operation, the set will always chose B as primary. If B becomes unavailable, the set will elect A as primary. For more information, see the Member Priority (page 390) documentation.
Data-Center Awareness
You can now tag replica set members to indicate their location. You can use these tags to design custom write rules (page 398) across data centers, racks, specic servers, or any other architecture choice. For example, an administrator can dene rules such as very important write or customerData or audit-trail to replicate to certain servers, racks, data centers, etc. Then in the application code, the developer would say:
db.foo.insert(doc, {w : "very important write"})
which would succeed if it fullled the conditions the DBA dened for very important write. For more information, see Tagging. Drivers may also support tag-aware reads. Instead of specifying slaveOk, you specify slaveOk with tags indicating which data-centers to read from. For details, see the MongoDB Drivers and Client Libraries (page 555) documentation.
1169
w : majority
You can also set w to majority to ensure that the write propagates to a majority of nodes, effectively committing it. The value for majority will automatically adjust as you add or remove nodes from the set. For more information, see Write Concern (page 398).
Reconguration with a Minority Up
If the majority of servers in a set has been permanently lost, you can now force a reconguration of the set to bring it back online. For more information see Recongure a Replica Set with Unavailable Members (page 458).
Primary Checks for a Caught up Secondary before Stepping Down
To minimize time without a primary, the rs.stepDown() (page 991) method will now fail if the primary does not see a secondary within 10 seconds of its latest optime. You can force the primary to step down anyway, but by default it will return an error message. See also Force a Member to Become Primary (page 437).
Extended Shutdown on the Primary to Minimize Interruption
When you call the shutdown (page 869) command, the primary will refuse to shut down unless there is a secondary whose optime is within 10 seconds of the primary. If such a secondary isnt available, the primary will step down and wait up to a minute for the secondary to be fully caught up before shutting down. Note that to get this behavior, you must issue the shutdown (page 869) command explicitly; sending a signal to the process will not trigger this behavior. You can also force the primary to shut down, even without an up-to-date secondary available.
Maintenance Mode
When repair or compact (page 859) runs on a secondary, the secondary will automatically drop into recovering mode until the operation nishes. This prevents clients from trying to read from it while its busy. Geospatial Features
Multi-Location Documents
Indexing is now supported on documents which have multiple location objects, embedded either inline or in nested sub-documents. Additional command options are also supported, allowing results to return with not only distance but the location used to generate the distance. For more information, see Multi-location Documents.
1170
Polygon searches
Polygonal $within (page 779) queries are also now supported for simple polygon shapes. For details, see the $within (page 779) operator documentation. Journaling Enhancements Journaling is now enabled by default for 64-bit platforms. Use the --nojournal command line option to disable it. The journal is now compressed for faster commits to disk. A new --journalCommitInterval (page 1025) run-time option exists for specifying your own group commit interval. The default settings do not change. A new { getLastError: { j: true } } (page 831) option is available to wait for the group commit. The group commit will happen sooner when a client is waiting on {j: true}. If journaling is disabled, {j: true} is a no-op. New ContinueOnError Option for Bulk Insert Set the continueOnError option for bulk inserts, in the driver (page 555), so that bulk insert will continue to insert any remaining documents even if an insert fails, as is the case with duplicate key exceptions or network interruptions. The getLastError (page 831) command will report whether any inserts have failed, not just the last one. If multiple errors occur, the client will only receive the most recent getLastError (page 831) results. See OP_INSERT. Note: For bulk inserts on sharded clusters, the getLastError (page 831) command alone is insufcient to verify success. Applications should must verify the success of bulk inserts in application logic.
Map Reduce
Output to a Sharded Collection
Using the new sharded ag, it is possible to send the result of a map/reduce to a sharded collection. Combined with the reduce or merge ags, it is possible to keep adding data to very large collections from map/reduce jobs. For more information, see MapReduce Output Options and mapReduce (page 814).
Performance Improvements
Map/reduce performance will benet from the following: Larger in-memory buffer sizes, reducing the amount of disk I/O needed during a job Larger javascript heap size, allowing for larger objects and less GC Supports pure JavaScript execution with the jsMode ag. See mapReduce (page 814).
1171
New Querying Features

Additional regex options: s
Allows the dot (.) to match all characters including new lines. This is in addition to the currently supported i, m and x. See Regular Expressions and $regex (page 774).
$and
A special boolean $and (page 768) query operator is now available. Command Output Changes The output of the validate (page 907) command and the documents in the system.profile collection have both been enhanced to return information as BSON objects with keys for each value rather than as free-form strings. Shell Features
Custom Prompt
You can dene a custom prompt for the mongo (page 1036) shell. You can change the prompt at any time by setting the prompt variable to a string or a custom JavaScript function returning a string. For examples, see Custom Prompt.
Default Shell Init Script
On startup, the shell will check for a .mongorc.js le in the users home directory. The shell will execute this le after connecting to the database and before displaying the prompt. If you would like the shell not to run the .mongorc.js le automatically, start the shell with --norc (page 1036). For more information, see mongo (page 1036). Most Commands Require Authentication In 2.0, when running with authentication (e.g. auth (page 1081)) all database commands require authentication, except the following commands. isMaster (page 838) authenticate getnonce buildInfo (page 871) ping (page 888) isdbgrid (page 848)
1172
68.2.3 Resources
MongoDB Downloads All JIRA Issues resolved in 2.0 All Backward Incompatible Changes

See the full index of this page for a complete list of changes included in 1.8. Upgrading (page 1173) Changes (page 1176) Resources (page 1178)
68.3.1 Upgrading
MongoDB 1.8 is a standard, incremental production release and works as a drop-in replacement for MongoDB 1.6, except: Replica set members should be upgraded in a particular order, as described in Upgrading a Replica Set (page 1173). The mapReduce (page 814) command has changed in 1.8, causing incompatibility with previous releases. mapReduce (page 814) no longer generates temporary collections (thus, keepTemp has been removed). Now, you must always supply a value for out. See the out eld options in the mapReduce (page 814) document. If you use MapReduce, this also likely means you need a recent version of your client driver. Preparation Read through all release notes before upgrading and ensure that no changes will affect your deployment. Upgrading a Standalone mongod 1. Download the v1.8.x binaries from the MongoDB Download Page. 2. Shutdown your mongod (page 1021) instance. 3. Replace the existing binary with the 1.8.x mongod (page 1021) binary. 4. Restart MongoDB. Upgrading a Replica Set 1.8.x secondaries can replicate from 1.6.x primaries. 1.6.x secondaries cannot replicate from 1.8.x primaries. Thus, to upgrade a replica set you must replace all of your secondaries rst, then the primary. For example, suppose you have a replica set with a primary, an arbiter and several secondaries. To upgrade the set, do the following: 68.3. Release Notes for MongoDB 1.8 1173
1. For the arbiter: (a) Shut down the arbiter. (b) Restart it with the 1.8.x binary from the MongoDB Download Page. 2. Change your cong (optional) to prevent election of a new primary. It is possible that, when you start shutting down members of the set, a new primary will be elected. To prevent this, you can give all of the secondaries a priority of 0 before upgrading, and then change them back afterwards. To do so: (a) Record your current cong. Run rs.config() (page 988) and paste the results into a text le. (b) Update your cong so that all secondaries have priority 0. For example:
config = rs.conf() { "_id" : "foo", "version" : 3, "members" : [ { "_id" : 0, "host" : "ubuntu:27017" }, { "_id" : 1, "host" : "ubuntu:27018" }, { "_id" : 2, "host" : "ubuntu:27019", "arbiterOnly" : true } { "_id" : 3, "host" : "ubuntu:27020" }, { "_id" : 4, "host" : "ubuntu:27021" }, ] } config.version++ 3 rs.isMaster() { "setName" : "foo", "ismaster" : false, "secondary" : true, "hosts" : [ "ubuntu:27017", "ubuntu:27018" ], "arbiters" : [ "ubuntu:27019" ], "primary" : "ubuntu:27018", "ok" : 1 }
1174
// for each secondary config.members[0].priority = 0 config.members[3].priority = 0 config.members[4].priority = 0 rs.reconfig(config)
3. For each secondary: (a) Shut down the secondary. (b) Restart it with the 1.8.x binary from the MongoDB Download Page. 4. If you changed the cong, change it back to its original state:
config = rs.conf() config.version++ config.members[0].priority = 1 config.members[3].priority = 1 config.members[4].priority = 1 rs.reconfig(config)
5. Shut down the primary (the nal 1.6 server), and then restart it with the 1.8.x binary from the MongoDB Download Page. Upgrading a Sharded Cluster 1. Turn off the balancer:
mongo <a_mongos_hostname> use config db.settings.update({_id:"balancer"},{$set : {stopped:true}}, true)
2. For each shard: If the shard is a replica set, follow the directions above for Upgrading a Replica Set (page 1173). If the shard is a single mongod (page 1021) process, shut it down and then restart it with the 1.8.x binary from the MongoDB Download Page. 3. For each mongos (page 1032): (a) Shut down the mongos (page 1032) process. (b) Restart it with the 1.8.x binary from the MongoDB Download Page. 4. For each cong server: (a) Shut down the cong server process. (b) Restart it with the 1.8.x binary from the MongoDB Download Page. 5. Turn on the balancer:
use config db.settings.update({_id:"balancer"},{$set : {stopped:false}})
Returning to 1.6 If for any reason you must move back to 1.6, follow the steps above in reverse. Please be careful that you have not inserted any documents larger than 4MB while running on 1.8 (where the max size has increased to 16MB). If you have you will get errors when the server tries to read those documents. 68.3. Release Notes for MongoDB 1.8 1175
Journaling
Returning to 1.6 after using 1.8 Journaling (page 71) works ne, as journaling does not change anything about the data le format. Suppose you are running 1.8.x with journaling enabled and you decide to switch back to 1.6. There are two scenarios: If you shut down cleanly with 1.8.x, just restart with the 1.6 mongod binary. If 1.8.x shut down uncleanly, start 1.8.x up again and let the journal les run to x any damage (incomplete writes) that may have existed at the crash. Then shut down 1.8.x cleanly and restart with the 1.6 mongod binary.
68.3.2 Changes
Journaling MongoDB now supports write-ahead Journaling (page 71) to facilitate fast crash recovery and durability in the storage engine. With journaling enabled, a mongod (page 1021) can be quickly restarted following a crash without needing to repair the collections. The aggregation framework makes it possible to do aggregation Sparse and Covered Indexes Sparse Indexes (page 334) are indexes that only include documents that contain the elds specied in the index. Documents missing the eld will not appear in the index at all. This can signicantly reduce index size for indexes of elds that contain only a subset of documents within a collection. Covered Indexes (page 342) enable MongoDB to answer queries entirely from the index when the query only selects elds that the index contains. Incremental MapReduce Support The mapReduce (page 814) command supports new options that enable incrementally updating existing collections. Previously, a MapReduce job could output either to a temporary collection or to a named permanent collection, which it would overwrite with new data. You now have several options for the output of your MapReduce jobs: You can merge MapReduce output into an existing collection. Output from the Reduce phase will replace existing keys in the output collection if it already exists. Other keys will remain in the collection. You can now re-reduce your output with the contents of an existing collection. Each key output by the reduce phase will be reduced with the existing document in the output collection. You can replace the existing output collection with the new results of the MapReduce job (equivalent to setting a permanent output collection in previous releases) You can compute MapReduce inline and return results to the caller without persisting the results of the job. This is similar to the temporary collections generated in previous releases, except results are limited to 8MB. For more information, see the out eld options in the mapReduce (page 814) document. Additional Changes and Enhancements
1.8.1
Sharding migrate x when moving larger chunks. 1176 Chapter 68. Previous Stable Releases
Durability x with background indexing. Fixed mongos concurrency issue with many incoming connections.
1.8.0
All changes from 1.7.x series.

1.7.6
Bug xes.
1.7.5
Journaling (page 71). Extent allocation improvements. Improved replica set connectivity for mongos (page 1032). getLastError (page 831) improvements for sharding.
1.7.4
mongos (page 1032) routes slaveOk queries to secondaries in replica sets. New mapReduce (page 814) output options. Sparse Indexes (page 334).
1.7.3
Initial covered index (page 342) support. Distinct can use data from indexes when possible. mapReduce (page 814) can merge or reduce results into an existing collection. mongod (page 1021) tracks and mongostat (page 1063) displays network usage. See mongostat (page 1063). Sharding stability improvements.
1.7.2
$rename (page 784) operator allows renaming of elds in a document. db.eval() (page 975) not to block. Geo queries with sharding. mongostat --discover (page 1065) option Chunk splitting enhancements. Replica sets network enhancements for servers behind a nat.
1177
1.7.1
Many sharding performance enhancements. Better support for $elemMatch (page 798) on primitives in embedded arrays. Query optimizer enhancements on range queries. Window service enhancements. Replica set setup improvements. $pull (page 791) works on primitives in arrays.
1.7.0
Sharding performance improvements for heavy insert loads. Slave delay support for replica sets. getLastErrorDefaults (page 466) for replica sets. Auto completion in the shell. Spherical distance for geo search. All xes from 1.6.1 and 1.6.2. Release Announcement Forum Pages 1.8.1, 1.8.0 1.7.6, 1.7.5, 1.7.4, 1.7.3, 1.7.2, 1.7.1, 1.7.0
68.3.3 Resources
MongoDB Downloads All JIRA Issues resolved in 1.8

See the full index of this page for a complete list of changes included in 1.6. Upgrading (page 1179) Sharding (page 1179) Replica Sets (page 1179) Other Improvements (page 1179) Installation (page 1179) 1.6.x Release Notes (page 1180) 1.5.x Release Notes (page 1180)
1178
68.4.1 Upgrading
MongoDB 1.6 is a drop-in replacement for 1.4. To upgrade, simply shutdown mongod (page 1021) then restart with the new binaries. Please note that you should upgrade to the latest version of whichever driver youre using. Certain drivers, including the Ruby driver, will require the upgrade, and all the drivers will provide extra features for connecting to replica sets.
68.4.2 Sharding
Sharding (page 483) is now production-ready, making MongoDB horizontally scalable, with no single point of failure. A single instance of mongod (page 1021) can now be upgraded to a distributed cluster with zero downtime when the need arises. Sharding (page 483) Deploy a Sharded Cluster (page 503) Convert a Replica Set to a Replicated Sharded Cluster (page 511)
68.4.3 Replica Sets

Replica sets (page 385), which provide automated failover among a cluster of n nodes, are also now available. Please note that replica pairs are now deprecated; we strongly recommend that replica pair users upgrade to replica sets. Replication (page 385) Deploy a Replica Set (page 419) Convert a Standalone to a Replica Set (page 423)
68.4.4 Other Improvements

The w option (and wtimeout) forces writes to be propagated to n servers before returning success (this works especially well with replica sets) $or queries (page 770) Improved concurrency $slice (page 801) operator for returning subsets of arrays 64 indexes per collection (formerly 40 indexes per collection) 64-bit integers can now be represented in the shell using NumberLong The findAndModify (page 826) command now supports upserts. It also allows you to specify elds to return $showDiskLoc option to see disk location of a document Support for IPv6 and UNIX domain sockets
68.4.5 Installation
Windows service improvements The C++ client is a separate tarball from the binaries 68.4. Release Notes for MongoDB 1.6 1179
68.4.6 1.6.x Release Notes

1.6.5
68.4.7 1.5.x Release Notes

1.5.8 1.5.7 1.5.6 1.5.5 1.5.4 1.5.3 1.5.2 1.5.1 1.5.0 You can see a full list of all changes on JIRA. Thank you everyone for your support and suggestions!

See the full index of this page for a complete list of changes included in 1.4. Upgrading (page 1180) Core Server Enhancements (page 1180) Replication and Sharding (page 1181) Deployment and Production (page 1181) Query Language Improvements (page 1181) Geo (page 1181)
68.5.1 Upgrading
Were pleased to announce the 1.4 release of MongoDB. 1.4 is a drop-in replacement for 1.2. To upgrade you just need to shutdown mongod (page 1021), then restart with the new binaries. (Users upgrading from release 1.0 should review the 1.2 release notes (page 1182), in particular the instructions for upgrading the DB format.) Release 1.4 includes the following improvements over release 1.2:
68.5.2 Core Server Enhancements

concurrency (page 727) improvements indexing memory improvements background index creation (page 336)
1180
better detection of regular expressions so the index can be used in more cases
68.5.3 Replication and Sharding

better handling for restarting slaves ofine for a while fast new slaves from snapshots (--fastsync) congurable slave delay (--slavedelay) replication handles clock skew on master $inc (page 784) replication xes sharding alpha 3 - notably 2-phase commit on cong servers
68.5.4 Deployment and Production

congure slow threshold for proling (page 99) ability to do fsync + lock (page 865) for backing up raw les option for separate directory per db (--directoryperdb) http://localhost:28017/_status to get serverStatus via http REST interface is off by default for security (--rest to enable) can rotate logs with a db command, logRotate (page 866) enhancements to serverStatus (page 889) command (db.serverStatus()) - counters and replication lag (page 444) stats new mongostat (page 1063) tool
68.5.5 Query Language Improvements

$all (page 763) with regex $not (page 770) partial matching of array elements $elemMatch (page 798) $ operator for updating arrays $addToSet (page 789) $unset (page 788) $pull (page 791) supports object matching $set (page 788) with array indices
68.5.6 Geo
2d geospatial search (page 369) geo $center and $box searches
1181
68.6 Release Notes for MongoDB 1.2.x

See the full index of this page for a complete list of changes included in 1.2. New Features (page 1182) DB Upgrade Required (page 1182) Replication Changes (page 1182) mongoimport (page 1182) eld lter changing (page 1183)
68.6.1 New Features

More indexes per collection Faster index creation Map/Reduce Stored JavaScript functions Congurable fsync time Several small features and xes
68.6.2 DB Upgrade Required

There are some changes that will require doing an upgrade if your previous version is <= 1.0.x. If youre already using a version >= 1.1.x then these changes arent required. There are 2 ways to do it: --upgrade stop your mongod (page 1021) process run ./mongod --upgrade start mongod (page 1021) again use a slave start a slave on a different port and data directory when its synced, shut down the master, and start the new slave on the regular port. Ask in the forums or IRC for more help.
68.6.3 Replication Changes

There have been minor changes in replication. If you are upgrading a master/slave setup from <= 1.1.2 you have to update the slave rst.
68.6.4 mongoimport
mongoimportjson has been removed and is replaced with mongoimport (page 1055) that can do json/csv/tsv
1182
68.6.5 eld lter changing

Weve changed the semantics of the eld lter a little bit. Previously only objects with those elds would be returned. Now the eld lter only changes the output, not which objects are returned. If you need that behavior, you can use $exists (page 772)
68.6. Release Notes for MongoDB 1.2.x
1183
1184
CHAPTER 69
Other MongoDB Release Notes
69.1 Default Write Concern Change

These release notes outline a change to all driver interfaces released in November 2012. See release notes for specic drivers for additional information.
69.1.1 Changes
As of the releases listed below, there are two major changes to all drivers: 1. All drivers will add a new top-level connection class that will increase consistency for all MongoDB client interfaces. This change is non-backward breaking: existing connection classes will remain in all drivers for a time, and will continue to operate as expected. However, those previous connection classes are now deprecated as of these releases, and will eventually be removed from the driver interfaces. The new top-level connection class is named MongoClient, or similar depending on how host languages handle namespacing. 2. The default write concern on the new MongoClient class will be to acknowledge all write operations 1 . This will allow your application to receive acknowledgment of all write operations. See the documentation of Write Concern (page 398) for more information about write concern in MongoDB. Please migrate to the new MongoClient class expeditiously.
69.1.2 Releases
The following driver releases will include the changes outlined in Changes (page 1185). See each drivers release notes for a full account of each release as well as other related driver-specic changes. C#, version 1.7 Java, version 2.10.0 Node.js, version 1.2 Perl, version 0.501.1
1 The drivers will call getLastError (page 831) without arguments, which is logically equivalent to the w: 1 option; however, this operation allows replica set users to override the default write concern with the getLastErrorDefaults (page 466) setting in the Replica Set Conguration (page 463).
1185
PHP, version 1.4 Python, version 2.4 Ruby, version 1.8
1186
Chapter 69. Other MongoDB Release Notes
CHAPTER 70
Version Numbers
There are three numbers in a MongoDB version: <major version number>.<release number>.<revision number> The second number, the release number, indicates release stability. An odd release number indicates a release in a development series. Even numbered releases indicate a stable, general availability, release. Additionally, the three numbers indicate the following: <major version number>: This rarely changes. A change to this number indicates very large changes to MongoDB. <release number>: A release can include many changes, including new features and updates. Some changes may break backwards compatibility. See release notes for every version for compatibility notes. Evennumbered release numbers are stable branches. Odd-numbered release numbers are development branches. <revision number>: This digit increments for every release. Changes each revision address bugs and security issues, Example Version numbers: 1.0.0 : First stable release 1.0.x : Bug xes to 1.0.x. These releases carry low risk. Always upgrade to the latest revision in your release series. 1.1.x : Development release. Includes new features not fully nished and other works-in-progress. Some things may be different than 1.0 1.2.x : Second stable release. This is a culmination of the 1.1.x development series.
1187
1188
Chapter 70. Version Numbers
Part XVI
About MongoDB Documentation
1189
The MongoDB Manual contains comprehensive documentation on the MongoDB document-oriented database management system. This page describes the manuals licensing, editions, and versions, and describes how to make a change request and how to contribute to the manual. For more information on MongoDB, see MongoDB: A Document Oriented Database. To download MongoDB, see the downloads page.
1191
1192
CHAPTER 71
License
This manual is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported (i.e. CC-BY-NC-SA) license. The MongoDB Manual is copyright 2011-2013 10gen, Inc.
1193
1194
Chapter 71. License
CHAPTER 72
Editions
In addition to the MongoDB Manual, you can also access this content in the following editions: ePub Format Single HTML Page PDF Format You also can access PDF les that contain subsets of the MongoDB Manual: MongoDB Reference Manual MongoDB Use Case Guide MongoDB CRUD Operation Introduction For Emacs users Info/Texinfo users, the following experimental Texinfo manuals are available for ofine use: MongoDB Manual Texinfo (tar.gz) MongoDB Reference Manual (tar.gz) MongoDB CRUD Operation Introduction (tar.gz) Important: The texinfo manuals are experimental. If you nd an issue with one of these editions, please le an issue in the DOCS Jira project.
1195
1196
Chapter 72. Editions
CHAPTER 73
Version and Revisions
This version of the manual reects version 2.4 of MongoDB. See the MongoDB Documentation Project Page for an overview of all editions and output formats of the MongoDB Manual. You can see the full revision history and track ongoing improvements and additions for all versions of the manual from its GitHub repository. This edition reects master branch of the documentation as of the 52b9250c0f087f8e354f0a8c1f1f3a845071ddb9 revision. This branch is explicitly accessible via http://docs.mongodb.org/master and you can always reference the commit of the current manual in the release.txt le. The most up-to-date, current, http://docs.mongodb.org/manual/. and stable version of the manual is always available at
1197
1198
Chapter 73. Version and Revisions
CHAPTER 74
Report an Issue or Make a Change Request
To report an issue with this manual or to make a change request, le a ticket at the MongoDB DOCS Project on Jira.
1199
1200
Chapter 74. Report an Issue or Make a Change Request
CHAPTER 75
Contribute to the Documentation
75.1 MongoDB Manual Translation

The original authorship language for all MongoDB documentation is American English. However, ensuring that speakers of other languages can read and understand the documentation is of critical importance to the documentation project. In this direction, the MongoDB Documentation project uses the service provided by Smartling to translate the MongoDB documentation into additional non-English languages. This translation project is largely supported by the work of volunteer translators from the MongoDB community who contribute to the translation effort. If you would like to volunteer to help translate the MongoDB documentation, please: complete the 10gen/MongoDB Contributor Agreement, and create an account on Smartling at translate.docs.mongodb.org. Please use the same email address you use to sign the contributor as you use to create your Smartling account. The mongodb-translators user group exists to facilitate collaboration between translators and the documentation team at large. You can join the Google Group without signing the contributors agreement. We currently have the following languages congured: Arabic Chinese Czech French German Hungarian Indonesian Italian Japanese Korean Lithuanian Polish Portuguese
1201
Romanian Russian Spanish Turkish Ukrainian If you would like to initiate a translation project to an additional language, please report this issue using the Report a Problem link above or by posting to the mongodb-translators list. Currently the translation project only publishes rendered translation. While the translation effort is currently focused on the web site we are evaluating how to retrieve the translated phrases for use in other media. See also: Contribute to the Documentation (page 1201) MongoDB Documentation Style and Conventions (page 1202) MongoDB Documentation Organization (page 1210) MongoDB Documentation Practices and Processes (page 1207) Build and Deploy the MongoDB Documentation (page 1211) The entire documentation source for this manual is available in the mongodb/docs repository, which is one of the MongoDB project repositories on GitHub. To contribute to the documentation, you can open a GitHub account, fork the mongodb/docs repository, make a change, and issue a pull request. In order for the documentation team to accept your change, you must complete the MongoDB/10gen Contributor Agreement. You can clone the repository by issuing the following command at your system shell:
git clone git://github.com/mongodb/docs.git
75.2 About the Documentation Process

The MongoDB Manual uses Sphinx, a sophisticated documentation engine built upon Python Docutils. The original reStructured Text les, as well as all necessary Sphinx extensions and build tools, are available in the same repository as the documentation. For more information on the MongoDB documentation process, see:
75.2.1 MongoDB Documentation Style and Conventions

This document provides an overview of the style for the MongoDB documentation stored in this repository. The overarching goal of this style guide is to provide an accessible base style to ensure that our documentation is easy to read, simple to use, and straightforward to maintain. For information regarding the MongoDB Manual organization, see MongoDB Documentation Organization (page 1210).
1202
Chapter 75. Contribute to the Documentation
Document History 2011-09-27: Document created with a (very) rough list of style guidelines, conventions, and questions. 2012-01-12: Document revised based on slight shifts in practice, and as part of an effort of making it easier for people outside of the documentation team to contribute to documentation. 2012-03-21: Merged in content from the Jargon, and cleaned up style in light of recent experiences. 2012-08-10: Addition to the Referencing section. 2013-02-07: Migrated this document to the manual. Added map-reduce terminology convention. Other edits. Naming Conventions This section contains guidelines on naming les, sections, documents and other document elements. File naming Convention: For Sphinx, all les should have a .txt extension. Separate words in le names with hyphens (i.e. -.) For most documents, le names should have a terse one or two word name that scribes the material covered in the document. Allow the path of the le within the ument tree to add some of the required context/categorization. For example its ceptable to have http://docs.mongodb.org/manual/core/sharding.rst http://docs.mongodb.org/manual/administration/sharding.rst. dedocacand
For tutorials, the full title of the document should be in the le name. For example, http://docs.mongodb.org/manual/tutorial/replace-one-configuration-server-in-a-shard Phrase headlines and titles so that they the content contained within the section so that users can determine what questions the text will answer, and material that it will address without needing them to read the content. This shortens the amount of time that people spend looking for answers, and improvise search/scanning, and possibly SEO. Prefer titles and headers in the form of Using foo over How to Foo. When using target references (i.e. :ref: references in documents,) use names that include enough context to be intelligible thought all documentations. For example, use replica-set-secondary-only-node as opposed to secondary-only-node. This is to make the source more usable and easier to maintain. Style Guide This includes the local typesetting, English, grammatical, conventions and preferences that all documents in the manual should use. The goal here is to choose good standards, that are clear, and have a stylistic minimalism that does not interfere with or distract from the content. A uniform style will improve user experience, and minimize the effect of a multi-authored document.
Punctuation
Use the oxford comma. Oxford commas are the commas in a list of things (e.g. something, something else, and another thing) before the conjunction (e.g. and or or.). Do not add two spaces after terminal punctuation, such as periods.
75.2. About the Documentation Process
1203
Place commas and periods inside quotation marks. Use title case for headings and document titles. Title case capitalizes the rst letter of the rst, last, and all signicant words.
Verbs
Verb tense and mood preferences, with examples: Avoid the rst person. For example do not say, We will begin the backup process by locking the database, or I begin the backup process by locking my database instance, Use the second person. If you need to back up your database, start by locking the database rst. In practice, however, its more concise to imply second person using the imperative, as in Before inititating a back up, lock the database. When indicated, use the imperative mood. For example: Backup your databases often and To prevent data loss, back up your databases. The future perfect is also useful in some cases. For example, Creating disk snapshots without locking the database will lead to an inconsistent state. Avoid helper verbs, as possible, to increase clarity and concision. For example, attempt to avoid this does foo and this will do foo when possible. Use does foo over will do foo in situations where this foos is unacceptable.
Referencing
To refer to future or planned functionality in MongoDB or a driver, always link to the Jira case. The Manuals conf.py provides an :issue: role that links directly to a Jira case (e.g. :issue:\SERVER-9001\). For non-object references (i.e. functions, operators, methods, database commands, settings) always reference only the rst occurrence of the reference in a section. You should always reference objects, except in section headings. Structure references with the why rst; the link second. For example, instead of this: Use the Convert a Replica Set to a Replicated Sharded Cluster (page 511) procedure if you have an existing replica set. Type this: To deploy a sharded cluster for an existing replica set, see Convert a Replica Set to a Replicated Sharded Cluster (page 511).
General Formulations
Contractions are acceptable insofar as they are necessary to increase readability and ow. Avoid otherwise. Make lists grammatically correct. Do not use a period after every item unless the list item completes the unnished sentence before the list. Use appropriate commas and conjunctions in the list items. Typically begin a bulleted list with an introductory sentence or clause, with a colon or comma. The following terms are one word: 1204 Chapter 75. Contribute to the Documentation
standalone workow Use unavailable, ofine, or unreachable to refer to a mongod instance that cannot be accessed. Do not use the colloquialism down. Always write out units (e.g. megabytes) rather than using abbreviations (e.g. MB.)
Structural Formulations
There should be at least two headings at every nesting level. Within an h2 block, there should be either: no h3 blocks, 2 h3 blocks, or more than 2 h3 blocks. Section headers are in title case (capitalize rst, last, and all important words) and should effectively describe the contents of the section. In a single document you should strive to have section titles that are not redundant and grammatically consistent with each other. Use paragraphs and paragraph breaks to increase clarity and ow. Avoid burying critical information in the middle of long paragraphs. Err on the side of shorter paragraphs. Prefer shorter sentences to longer sentences. Use complex formations only as a last resort, if at all (e.g. compound complex structures that require semi-colons). Avoid paragraphs that consist of single sentences as they often represent a sentence that has unintentionally become too complex or incomplete. However, sometimes such paragraphs are useful for emphasis, summary, or introductions. As a corollary, most sections should have multiple paragraphs. For longer lists and more complex lists, use bulleted items rather than integrating them inline into a sentence. Do not expect that the content of any example (inline or blocked,) will be self explanatory. Even when it feels redundant, make sure that the function and use of every example is clearly described.
ReStructured Text and Typesetting
Place spaces between nested parentheticals and elements in JavaScript examples. For example, prefer { [ a, a, a ] } over {[a,a,a]}. For underlines associated with headers in RST, use: = for heading level 1 or h1s. Use underlines and overlines for document titles. - for heading level 2 or h2s. ~ for heading level 3 or h3s. for heading level 4 or h4s. Use hyphens (-) to indicate items of an ordered list. Place footnotes and other references, if you use them, at the end of a section rather than the end of a le. Use the footnote format that includes automatic numbering and a target name for ease of use. For instance a footnote tag may look like: [#note]_ with the corresponding directive holding the body of the footnote that resembles the following: .. [#note]. Do not include .. code-block:: [language] in footnotes.
1205
As it makes sense, use the .. code-block:: [language] form to insert literal blocks into the text. While the double colon, ::, is functional, the .. code-block:: [language] form makes the source easier to read and understand. For all mentions of referenced types (i.e. commands, operators, expressions, functions, statuses, etc.) use the reference types to ensure uniform formatting and cross-referencing. Jargon and Common Terms
Database Systems and Processes
To indicate the entire database system, use MongoDB, not mongo or Mongo. To indicate the database process or a server instance, use mongod or mongos. Refer to these as processes or instances. Reserve database for referring to a database structure, i.e., the structure that holds collections and refers to a group of les on disk.
Distributed System Terms
Refer to partitioned systems as sharded clusters. Do not use shard clusters or sharded systems. Refer to congurations that run with replication as replica sets (or master/slave deployments) rather than clusters or other variants.
Data Structure Terms
document refers to rows or records in a MongoDB database. Potential confusion with JSON Documents. Do not refer to documents as objects, because drivers (and MongoDB) do not preserve the order of elds when fetching data. If the order of objects matter, use an array. eld refers to a key or identier of data within a MongoDB document. value refers to the contents of a eld. sub-document describes a nested document.
Other Terms
Use example.net (and .org or .com if needed) for all examples and samples. Hyphenate map-reduce in order to avoid ambiguous reference to the command name. Do not camel-case. Notes on Specic Features Geo-Location 1. While MongoDB is capable of storing coordinates in sub-documents, in practice, users should only store coordinates in arrays. (See: DOCS-41.)
1206
75.2.2 MongoDB Documentation Practices and Processes

This document provides an overview of the practices and processes. Contents MongoDB Documentation Practices and Processes (page 1207) Commits (page 1207) Standards and Practices (page 1207) Collaboration (page 1207) Builds (page 1208) Publication (page 1208) Branches (page 1208) Migration from Legacy Documentation (page 1208) Review Process (page 1209) * Types of Review (page 1209) Initial Technical Review (page 1209) Content Review (page 1209) Consistency Review (page 1209) Subsequent Technical Review (page 1209) Review Methods (page 1209) *
Commits When relevant, include a Jira case identier in a commit message. Reference documentation cases when applicable, but feel free to reference other cases from jira.mongodb.org. Err on the side of creating a larger number of discrete commits rather than bundling large set of changes into one commit. For the sake of consistency, remove trailing whitespaces in the source le. Hard wrap les to between 72 and 80 characters per-line. Standards and Practices At least two people should vet all non-trivial changes to the documentation before publication. One of the reviewers should have signicant technical experience with the material covered in the documentation. All development and editorial work should transpire on github branches or forks that editors can then merge into the publication branches. Collaboration To propose a change to the documentation, do either of the following: Open a ticket in the documentation project proposing the change. Someone on the documentation team will make the change and be in contact with you so that you can review the change. Using GitHub, fork the mongodb/docs repository, commit your changes, and issue a pull request. Someone on the documentation team will review and incorporate your change into the documentation.
1207
Builds Building the documentation is useful because Sphinx and docutils can catch numerous errors in the format and syntax of the documentation. Additionally, having access to an example documentation as it will appear to the users is useful for providing more effective basis for the review process. Besides Sphinx, Pygments, and Python-Docutils, the documentation repository contains all requirements for building the documentation resource. Talk to someone on the documentation team if you are having problems running builds yourself. Publication The makele for this repository contains targets that automate the publication process. Use make html to publish a test build of the documentation in the build/ directory of your repository. Use make publish to build the full contents of the manual from the current branch in the ../public-docs/ directory relative the docs repository. Other targets include: man - builds UNIX Manual pages for all Mongodb utilities. push - builds and deploys the contents of the ../public-docs/. pdfs - builds a PDF version of the manual (requires LaTeX dependencies.) Branches This section provides an overview of the git branches in the MongoDB documentation repository and their use. At the present time, future work transpires in the master, with the main publication being current. As the documentation stabilizes, the documentation team will begin to maintain branches of the documentation for specic MongoDB releases. Migration from Legacy Documentation The MongoDB.org Wiki contains a wealth of information. As the transition to the Manual (i.e. this project and resource) continues, its critical that no information disappears or goes missing. The following process outlines how to migrate a wiki page to the manual: 1. Read the relevant sections of the Manual, and see what the new documentation has to offer on a specic topic. In this process you should follow cross references and gain an understanding of both the underlying information and how the parts of the new content relates its constituent parts. 2. Read the wiki page you wish to redirect, and take note of all of the factual assertions, examples presented by the wiki page. 3. Test the factual assertions of the wiki page to the greatest extent possible. Ensure that example output is accurate. In the case of commands and reference material, make sure that documented options are accurate. 4. Make corrections to the manual page or pages to reect any missing pieces of information. The target of the redirect need not contain every piece of information on the wiki page, if the manual as a whole does, and relevant section(s) with the information from the wiki page are accessible from the target of the redirection. 5. As necessary, get these changes reviewed by another writer and/or someone familiar with the area of the information in question. At this point, update the relevant Jira case with the target that youve chosen for the redirect, and make the ticket unassigned. 1208 Chapter 75. Contribute to the Documentation
6. When someone has reviewed the changes and published those changes to Manual, you, or preferably someone else on the team, should make a nal pass at both pages with fresh eyes and then make the redirect. Steps 1-5 should ensure that no information is lost in the migration, and that the nal review in step 6 should be trivial to complete. Review Process
Types of Review
The content in the Manual undergoes many types of review, including the following: Initial Technical Review Review by an engineer familiar with MongoDB and the topic area of the documentation. This review focuses on technical content, and correctness of the procedures and facts presented, but can improve any aspect of the documentation that may still be lacking. When both the initial technical review and the content review are complete, the piece may be published. Content Review Textual review by another writer to ensure stylistic consistency with the rest of the manual. Depending on the content, this may precede or follow the initial technical review. When both the initial technical review and the content review are complete, the piece may be published. Consistency Review This occurs post-publication and is content focused. The goals of consistency reviews are to increase the internal consistency of the documentation as a whole. Insert relevant cross-references, update the style as needed, and provide background fact-checking. When possible, consistency reviews should be as systematic as possible and we should avoid encouraging stylistic and information drift by editing only small sections at a time. Subsequent Technical Review If the documentation needs to be updated following a change in functionality of the server or following the resolution of a user issue, changes may be signicant enough to warrant additional technical review. These reviews follow the same form as the initial technical review, but is often less involved and covers a smaller area.
Review Methods
If youre not a usual contributor to the documentation and would like to review something, you can submit reviews in any of the following methods: If youre reviewing an open pull request in GitHub, the best way to comment is on the overview diff, which you can nd by clicking on the diff button in the upper left portion of the screen. You can also use the following URL to reach this interface:
https://github.com/mongodb/docs/pull/[pull-request-id]/files
Replace [pull-request-id] with the identier of the pull request. Make all comments inline, using GitHubs comment system. You may also provide comments directly on commits, or on the pull request itself but these commit-comments are archived in less coherent ways and generate less useful emails, while comments on the pull request lead to less specic changes to the document. Leave feedback on Jira cases in the DOCS project. These are better for more general changes that arent necessarily tied to a specic line, or affect multiple les. 75.2. About the Documentation Process 1209
Create a fork of the repository in your GitHub account, make any required changes and then create a pull request with your changes. If you insert lines that begin with any of the following annotations:
.. TODO: TODO: .. TODO TODO
followed by your comments, it will be easier for the original writer to locate your comments. The two dots .. format is a comment in reStructured Text, which will hide your comments from Sphinx and publication if youre worried about that. This format is often easier for reviewers with larger portions of content to review.
75.2.3 MongoDB Documentation Organization

This document provides an overview of the global organization of the documentation resource. Refer to the notes below if you are having trouble understanding the reasoning behind a les current location, or if you want to add new documentation but arent sure how to integrate it into the existing resource. If you have questions, dont hesitate to open a ticket in the Documentation Jira Project or contact the documentation team. Global Organization
Indexes and Experience
The documentation project has two index les: http://docs.mongodb.org/manual/contents.txt and http://docs.mongodb.org/manual/index.txt. The contents le provides the documentations tree structure, which Sphinx uses to create the left-pane navigational structure, to power the Next and Previous page functionality, and to provide all overarching outlines of the resource. The index le is not included in the contents le (and thus builds will produce a warning here) and is the page that users rst land on when visiting the resource. Having separate contents and index les provides a bit more exibility with the organization of the resource while also making it possible to customize the primary user experience. Additionally, in the top level of the source/ directory, there are a number of topical index or outline les. These (like the index and contents les) use the .. toctree:: directive to provide organization within the documentation. The subject-specic landing pages indexes combine to create the index in the contents le.
Topical Indexes and Meta Organization
Because the documentation on any given subject exists in a number of different locations across the resource the topical indexes provide the real structure and organization to the resource. This organization makes it possible to provide great exibility while still maintaining a reasonable organization of les and URLs for the documentation. Consider the following example: Given that topic such as replication, has material regarding the administration of replica sets, as well as reference material, an overview of the functionality, and operational tutorials, it makes more sense to include a few locations for documents, and use the meta documents to provide the topic-level organization. Current landing pages include:
1210
administration applications crud faq mongo reference replication security sharding Additional topical indexes are forthcoming. The Top Level Folders The documentation has a number of top-level folders, that hold all of the content of the resource. Consider the following list and explanations below: administration - contains all of the operational and architectural information that systems and database administrators need to know in order to run MongoDB. Topics include: monitoring, replica sets, shard clusters, deployment architectures, and conguration. applications - contains information about application development and use. While most documentation regarding application development is within the purview of the driver documentation, there are some larger topics regarding the use of these features that deserve some coverage in this context. Topics include: drivers, schema design, optimization, replication, and sharding. applications/use-cases - contains use cases that detail how MongoDB can support various kinds uses and application designs, including in depth instructions and examples. core - contains overviews and introduction to the core features, functionality, and concepts of MongoDB. Topics include: replication, sharding, capped collections, journaling/durability, aggregation. reference - contains references and indexes of shell functions, database commands, status outputs, as well as manual pages for all of the programs come with MongoDB (e.g. mongostat and mongodump.) tutorial - contains operational guides and tutorials that lead users through common tasks (administrative and conceptual) with MongoDB. This includes programming patterns and operational guides. faq - contains all the frequently asked questions related to MongoDB, in a collection of topical les.
75.2.4 Build and Deploy the MongoDB Documentation

This document contains more direct instructions for building the MongoDB documentation. Requirements For basic publication and testing: GNU Make Python Git
1211
Sphinx (documentation management toolchain) Pygments (syntax highlighting) PyYAML (for the generated tables) For full publication builds: python-argparse LaTeX/PDF LaTeX (typically texlive; for building PDFs) Common Utilities (rsync, tar, gzip, sed) Building the Documentation Clone the repository:
git clone git://github.com/mongodb/docs.git
To build the full publication version of the manual, you will need to have a function LaTeX tool chain; however, for routine day-to-day rendering of the documentation you can install a much more minimal tool chain.
For Routine Builds
Begin by installing dependencies. On Arch Linux, use the following command to install the full dependencies:
pacman -S python2-sphinx python2-pygments python2-yaml
On Debian/Ubuntu systems issue the following command:

apt-get install python-sphinx python-yaml python-argparse
To build the documentation issue the following command:

make html
You can nd the build output in build/<branch>/html, where <branch> is the name of your current branch.
For Publication Builds
Begin by installing additional dependencies. On Arch Linux, use the following command to install the full dependencies:
pacman -S python2-sphinx python2-pygments python2-yaml \ texlive-bin texlive-core texlive-latexextra
On Debian/Ubuntu systems use the following command:

apt-get install python-yaml python-argparse python-sphinx \ texlive-latex-recommended texlive-latex-recommended
Note: The Debian/Ubuntu dependencies, have not been thoroughly tested. If you nd an additional dependency, please submit a pull request to modify this document. On OS X: 1. You may need to use easy_install to install pip using the following command if you have not already done so:
1212
easy_install pip
Alternately, you may be able to replace pip with easy_install in the next step. 2. Install Sphinx, Docutils, and their dependencies with pip using the following command:
pip install Sphinx Jinja2 Pygments docutils PyYAML
Jinja2, Pygments, and docutils are all dependencies of Sphinx. Note: As of June 6, 2012 and Sphinx version 1.1.3, you must compile the MongoDB documentation using the Python 2.x series version of Sphinx. There are serious generation problems with the Python 3 series version of Sphinx. 3. Install a TeX distribution (for building the PDF.) If you do not have a LaTeX installation, use MacTeX If you have any corrections to the instructions for these platforms or you have a dependency list for Fedora, CentOS, Red Hat, or other related distributions, please submit a pull request to add this information to this document. To build a test version of the Manual, issue the following command:
make publish
This places a complete version of the manual in ../public-docs/ named for the current branch (as of 201203-19, typically master.) To publish a new build of the manual, issue the following command:
make push
Warning: This target depends on publish, and simply uses rsync to move the content of the ../public-docs/ to the web servers. You must have the proper credentials to run these operations. Run publish procedure and thoroughly test the build before pushing it live. Troubleshooting If you encounter problems with the build, please contact the docs team, so that we can update this guide and/or x the build process. Build Components and Internals This section describes the build system for the MongoDB manual, including the custom components, and the organization of the production build process, and the implementation and encoding of the build process.
Tables
bin/table_builder.py provides a way to generate easy to main reStructuredText tables, from content stored in YAML les. Rationale: reStructuredTexts default tables are easy to read in source format, but expensive to maintain, particularly with larger numbers of columns, because changing widths of column necessitates reformatting the entire table. reStructuredText does provide a more simple list table format for simple tables, but these tables do not support more complex multi-line output. Solution: table_builder.py reads a .yaml le that contains three documents:
1213
(Each document has a section eld that holds the name/type of the section, that table_builder.py uses to ensure that the YAML le is well formed.) 1. A layout document that describes the structure the nal presentation of the table. Contains two eld, a header that holds a list of eld references, and a rows eld that holds a list of lists of eld references, for example:
section: layout header: [ meta.header1, meta.header2 ] rows: - 1: [ content.sql1, content.mongo1 ] - 2: [ content.sql2, content.mongo2 ] - 3: [ content.sql3, content.mongo3 ] - 4: [ content.sql4, content.mongo4 ]
2. A meta document that holds row, column or other minor descriptions, referenced in the layout section. 3. A content document that holds the major content of the document. There is no functional difference between meta and content elds except that they each provide a distinct namespace for table content. table_builder.py generates .rst output les from .yaml les. The documents processed by Sphinx use the .. include:: reStructureText directive to include the .rst le. The build system includes targets (generated,) for all tables, which are a dependency of the Sphinx build process. 1 Use: To add a table: create an appropriate .yaml le using any of the existing les as an example. The build system generates all table les in the source/includes/ directory with the .yaml extension that begin with table-. include the generated .rst le in your Sphinx document. (Optional.)
Generated Makeles
System While the makefile in the top level of documentation source coordinates the build process, most of the build targets and build system exist in the form of makeles generated by a collection of Python scripts. This architecture reduces redundancy while increasing clarity and consistency. These makeles enter the build process by way of include statements and a pattern rule in bin/makefile.dynamic, as follows:
-include $(output)/makefile.tables -include $(output)/makefile.sphinx $(output)/makefile.%:bin/makefile-builder/%.py bin/makefile_builder.py bin/builder_data.py @$(PYTHONBIN) bin/makefile-builder/$(subst .,,$(suffix $@)).py $@
This will rebuild any of the include les that match the pattern $(output)/makefile.%, if the corresponding python script changes, or it will rebuild all generated makeles if the builder_data.py or the makefile_builder.py les change. The Python scripts that output these makeles, all use the MakefileBuilder class in the makefile_builder.py le, and are all located in the bin/makefile-builder/ directory. Consider a simplied example Python code:
1 To prevent a build error, tables are a dependency of all Sphinx builds except the dirhtml, singlehtml, and latex builds, which run concurrently during the production build process. If you change tables, and run any of these targets without building the tables target, you the table will not refresh.
1214
from makefile_builder import MakefileBuilder from builder_data import sphinx m = MakefileBuilder() m.section_break(sphinx targets, block=sphinx) m.comment(each sphinx target invokes and controls the sphinx build., block=sphinx) m.newline(block=sphinx) for (builder, prod) in sphinx: m.newline(1, builder) m.append_var(sphinx-targets, builder) if prod is True and builder != epub: b = production m.target(builder, block=b) else: b = testing m.target(builder, sphinx-prerequisites, block=b) m.job(mkdir -p $(branch-output)/ + builder, block=b) m.msg([$@]: created $(branch-output)/ + builder, block=b) m.msg([sphinx]: starting $@ build, block=b) m.msg([$@]: build started at date., block=b) m.job($(SPHINXBUILD) -b $@ $(ALLSPHINXOPTS) $(branch-output)/$@, block=b) m.write(makefile.output-filename)
You can also call m.print_content() to render the makele to standard output. See makefile_builder.py for the more methods that you can use to dene makeles. This code will generate a makele that resembles the following:
sphinx-targets += epub epub:sphinx-prerequisites @mkdir -p $(branch-output)/epub @echo [$@]: created $(branch-output)/epub @echo [sphinx]: starting $@ build @echo [$@]: build started at date. @$(SPHINXBUILD) -b $@ $(ALLSPHINXOPTS) $(branch-output)/$@ sphinx-targets += html html:sphinx-prerequisites @mkdir -p $(branch-output)/html @echo [$@]: created $(branch-output)/html @echo [sphinx]: starting $@ build @echo [$@]: build started at date. @$(SPHINXBUILD) -b $@ $(ALLSPHINXOPTS) $(branch-output)/$@ sphinx-targets += gettext gettext:sphinx-prerequisites @mkdir -p $(branch-output)/gettext @echo [$@]: created $(branch-output)/gettext @echo [sphinx]: starting $@ build @echo [$@]: build started at date. @$(SPHINXBUILD) -b $@ $(ALLSPHINXOPTS) $(branch-output)/$@
All information about the targets themselves are in the builder_data.py le, that contains a number of variables that hold lists of tuples with information used by the Python scripts to generate the build rules. Comments explain the
1215
structure of the data in builder_data.py. System The build system contains the following 8 makeles: pdfs: Encodes the process for transforming Sphinxs LaTeX output into pdfs. tables: Describes the process for building all tables generated using table_builder.py. links: Creates the symbolic links required for production builds. sphinx: Generates the targets for Sphinx. These are mostly, but not entirely consistent with the default targets provided by Sphinx itself. releases: Describe targets for generating les for inclusion in the installation have the versions of MongoDB automatically baked into their text. errors: Special processing of the HTTP error pages. migrations: Describes the migration process for all non-sphinx components of the build. sphinx-migrations: Ensures that all sphinx migrations are fresh. Troubleshooting If you experience an issue with the generated makeles, the generated les have comments, and are quite human readable. To add new generated targets or makeles, experiment rst writing makeles themselves, and then write scripts to generate the makeles. Because the generated makeles, and indeed most of the build process does not echo commands, use make -n to determine the actual oration and sequence used in the build process. If you have any questions, please feel free to open a Jira Case.
1216

MongoDB Manual

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

MongoDB Manual

Încărcat de

Drepturi de autor:

Formate disponibile

MongoDB Documentation

MongoDB Documentation Project

May 09, 2013

Core MongoDB Operations (CRUD)

25.3 25.4 25.5 25.6 25.7 25.8 25.9 25.10

241 243 244 244 245 246 247 247

The mongo Shell

45 Getting Started with the mongo Shell vi

45.1 45.2 45.3 45.4 45.5 45.6

Frequently Asked Questions

57.4 57.5 57.6 57.7 57.8 57.9 57.10 57.11

66 General Reference 1105 66.1 MongoDB Limits and Thresholds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1105

66.2 66.3 66.4 66.5 66.6

1108 1113 1115 1117 1119

About MongoDB Documentation

1.1 Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux

1.1.2 Package Options

MongoDB Documentation, Release 2.4.3

1.1.3 Install MongoDB

Chapter 1. Installation Guides

MongoDB Documentation, Release 2.4.3

1.1.4 Congure MongoDB

1.1.5 Control MongoDB

1.1. Install MongoDB on Red Hat Enterprise, CentOS, or Fedora Linux

MongoDB Documentation, Release 2.4.3

1.1.6 Using MongoDB

1.2 Install MongoDB on Ubuntu

Chapter 1. Installation Guides

MongoDB Documentation, Release 2.4.3

1.2.2 Package Options

1.2.3 Install MongoDB

echo deb http://downloads-distro.mongodb.org/repo/ubuntu-upstart dist 10gen | sudo tee /etc/apt/sou

Now issue the following command to reload your repository:

1.2. Install MongoDB on Ubuntu

MongoDB Documentation, Release 2.4.3

echo "mongodb-10gen hold" | dpkg --set-selections

1.2.4 Congure MongoDB

1.2.5 Controlling MongoDB

Chapter 1. Installation Guides

MongoDB Documentation, Release 2.4.3

1.2.6 Using MongoDB

1.3 Install MongoDB on Debian

1.3.2 Package Options

1.3. Install MongoDB on Debian

MongoDB Documentation, Release 2.4.3

1.3.3 Install MongoDB

echo deb http://downloads-distro.mongodb.org/repo/debian-sysvinit dist 10gen | sudo tee /etc/apt/so

Now issue the following command to reload your repository:

1.3.4 Congure MongoDB

Chapter 1. Installation Guides

MongoDB Documentation, Release 2.4.3

1.3.5 Controlling MongoDB

1.3.6 Using MongoDB

1.4 Install MongoDB on Linux

MongoDB Documentation, Release 2.4.3

1.4.2 Download MongoDB

Chapter 1. Installation Guides

MongoDB Documentation, Release 2.4.3

chown mongo /data/db

1.5 Install MongoDB on OS X

MongoDB Documentation, Release 2.4.3

1.5.2 Install with Package Management

Chapter 1. Installation Guides

MongoDB Documentation, Release 2.4.3